Groups keyboard shortcuts have been updated
Dismiss
See shortcuts
veren...@gmail.com
Sep 18, 2018, 10:40:53 AM9/18/18
to lavaan
Hi all,
I have a problem with running a CFA with WLSMV (dichotomous data).
R has the basic setting of fixing the first factor loading of a factor. I want all factor loadings to be free estimated. How can I do this? The following syntax doesn't work
myModel11 <- "FA =~ NA* T1_OD1a + T1_OR1a + V1_OD1a + V1_OR1a + T2_OD1a + T2_OR1a + V2_OD1a + V2_OR1a"
fit11 <- cfa(myModel11, data=data.T1, estimator = "WLSMV", ordered = c("T1_OD1a, T1_OR1a, V1_OD1a, V1_OR1a, T2_OD1a, T2_OR1a, V2_OD1a, V2_OR1a"))
model <- summary(fit11, fit.measures = TRUE, rsq=TRUE, standardized=TRUE)
modificationIndices(fit11)
reliability(fit11)
fit11 <- cfa(myModel11, data=data.T1, estimator = "WLSMV", ordered = c("T1_OD1a, T1_OR1a, V1_OD1a, V1_OR1a, T2_OD1a, T2_OR1a, V2_OD1a, V2_OR1a"))
model <- summary(fit11, fit.measures = TRUE, rsq=TRUE, standardized=TRUE)
modificationIndices(fit11)
reliability(fit11)
Similar problem:
In another case I want to fix parameters. But this synatx doesn't work either:
myModel32 <- 'Video 1 =~ V1_OD1a + V1_OR1a
Video 2 =~ V2_OD1a + V2_OR1a
Text 1=~ T1_OD1a + T1_OR1a
Text 2=~ T2_OD1a + T2_OR1a
Diagnostik =~ T1_OD1a + 1*V1_OD1a + 1*T2_OD1a + 1*V2_OD1a
Rückmeldung =~ T1_OR1a + 1*V1_OR1a + 1*T2_OR1a + 1*V2_OR1a
'
fit32 <- cfa(myModel32, data = data.T1,estimator = "WLSMV", ordered = c("T1_OD1a, T1_OR1a, V1_OD1a, V1_OR1a, T2_OD1a, T2_OR1a, V2_OD1a, V2_OR1a"))
summary(fit32, fit.measures =T, standardized = T, rsq=T)
Video 2 =~ V2_OD1a + V2_OR1a
Text 1=~ T1_OD1a + T1_OR1a
Text 2=~ T2_OD1a + T2_OR1a
Diagnostik =~ T1_OD1a + 1*V1_OD1a + 1*T2_OD1a + 1*V2_OD1a
Rückmeldung =~ T1_OR1a + 1*V1_OR1a + 1*T2_OR1a + 1*V2_OR1a
'
fit32 <- cfa(myModel32, data = data.T1,estimator = "WLSMV", ordered = c("T1_OD1a, T1_OR1a, V1_OD1a, V1_OR1a, T2_OD1a, T2_OR1a, V2_OD1a, V2_OR1a"))
summary(fit32, fit.measures =T, standardized = T, rsq=T)
Thank you for your help!
Terrence Jorgensen
Sep 24, 2018, 5:29:54 AM9/24/18
to lavaan
I want all factor loadings to be free estimated. How can I do this? The following syntax doesn't work
Works fine for me.
In another case I want to fix parameters. But this synatx doesn't work either:
Works fine for me.
myModel32 <- 'Video 1 =~ NA*V1_OD1a + V1_OR1a
Video 2 =~ NA*V2_OD1a + V2_OR1a
Text 1=~ NA*T1_OD1a + T1_OR1a
Text 2=~ NA*T2_OD1a + T2_OR1a
Diagnostik =~ T1_OD1a + 1*V1_OD1a + 1*T2_OD1a + 1*V2_OD1a
Rückmeldung =~ T1_OR1a + 1*V1_OR1a + 1*T2_OR1a + 1*V2_OR1a
'
fit32 <- cfa(myModel32)
parTable(fit32)What version of lavan are you using?
sessionInfo()Terrence D. Jorgensen
Postdoctoral Researcher, Methods and Statistics
Research Institute for Child Development and Education, the University of Amsterdam
veren...@gmail.com
Sep 24, 2018, 7:09:04 AM9/24/18
to lavaan
I´m using lavaan_0.6-3.
Maybe the warning messages might help?:
Results using "NA*"
> myModel11 <- "FA =~ NA*T1_OD1a + T1_OR1a + V1_OD1a + V1_OR1a + T2_OD1a + T2_OR1a + V2_OD1a + V2_OR1a"
> fit11 <- cfa(myModel11, data=data.T1, estimator = "WLSMV", ordered = c("T1_OD1a, T1_OR1a, V1_OD1a, V1_OR1a, T2_OD1a, T2_OR1a, V2_OD1a, V2_OR1a"))
Warning messages:
1: In lav_partable_check(lavpartable, categorical = lavoptions$categorical, :
lavaan WARNING: parameter table does not contain thresholds
2: In lav_model_vcov(lavmodel = lavmodel, lavsamplestats = lavsamplestats, :
lavaan WARNING:
Could not compute standard errors! The information matrix could
not be inverted. This may be a symptom that the model is not
identified.
3: In lav_test_satorra_bentler(lavobject = NULL, lavsamplestats = lavsamplestats, :
lavaan WARNING: could not invert information matrix
> model <- summary(fit11, fit.measures = TRUE, rsq=TRUE, standardized=TRUE)
lavaan 0.6-3 ended normally after 32 iterations
Optimization method NLMINB
Number of free parameters 25
Used Total
Number of observations 276 277
Estimator DWLS
Model Fit Test Statistic 18.481
Degrees of freedom 19
P-value (Chi-square) 0.491
Error in TEST[[2]] : subscript out of boundsResults without "NA*":
> myModel11 <- "FA =~ T1_OD1a + T1_OR1a + V1_OD1a + V1_OR1a + T2_OD1a + T2_OR1a + V2_OD1a + V2_OR1a"
> fit11 <- cfa(myModel11, data=data.T1, estimator = "WLSMV", ordered = c("T1_OD1a, T1_OR1a, V1_OD1a, V1_OR1a, T2_OD1a, T2_OR1a, V2_OD1a, V2_OR1a"))
Warning messages:
1: In lav_partable_check(lavpartable, categorical = lavoptions$categorical, :
lavaan WARNING: parameter table does not contain thresholds
2: In lav_model_vcov(lavmodel = lavmodel, lavsamplestats = lavsamplestats, :
lavaan WARNING:
The variance-covariance matrix of the estimated parameters (vcov)
does not appear to be positive definite! The smallest eigenvalue
(= -5.703747e-18) is smaller than zero. This may be a symptom that
the model is not identified.
> model <- summary(fit11, fit.measures = TRUE, rsq=TRUE, standardized=TRUE)
lavaan 0.6-3 ended normally after 37 iterations
Optimization method NLMINB
Number of free parameters 24
Used Total
Number of observations 276 277
Estimator DWLS Robust
Model Fit Test Statistic 18.481 35.161
Degrees of freedom 20 20
P-value (Chi-square) 0.556 0.019
Scaling correction factor 0.568
Shift parameter 2.640
for simple second-order correction (Mplus variant)
Model test baseline model:
Minimum Function Test Statistic 257.648 136.904
Degrees of freedom 28 28
P-value 0.000 0.000
User model versus baseline model:
Comparative Fit Index (CFI) 1.000 0.861
Tucker-Lewis Index (TLI) 1.009 0.805
Robust Comparative Fit Index (CFI) NA
Robust Tucker-Lewis Index (TLI) NA
Root Mean Square Error of Approximation:
RMSEA 0.000 0.053
90 Percent Confidence Interval 0.000 0.048 0.021 0.081
P-value RMSEA <= 0.05 0.961 0.409
Robust RMSEA NA
90 Percent Confidence Interval NA NA
Standardized Root Mean Square Residual:
SRMR 0.055 0.055
Parameter Estimates:
Information Expected
Information saturated (h1) model Unstructured
Standard Errors Robust.sem
Latent Variables:
Estimate Std.Err z-value P(>|z|) Std.lv Std.all
FA =~
T1_OD1a 1.000 0.177 0.463
T1_OR1a 0.609 0.168 3.631 0.000 0.108 0.284
V1_OD1a 1.173 0.218 5.393 0.000 0.208 0.463
V1_OR1a 0.783 0.175 4.474 0.000 0.139 0.421
T2_OD1a 1.479 0.264 5.612 0.000 0.262 0.660
T2_OR1a 0.524 0.153 3.428 0.001 0.093 0.366
V2_OD1a 1.536 0.290 5.289 0.000 0.272 0.583
V2_OR1a 0.628 0.178 3.533 0.000 0.111 0.410
Intercepts:
Estimate Std.Err z-value P(>|z|) Std.lv Std.all
.T1_OD1a 1.178 0.023 51.102 0.000 1.178 3.076
.T1_OR1a 1.174 0.023 51.360 0.000 1.174 3.091
.V1_OD1a 1.279 0.027 47.290 0.000 1.279 2.847
.V1_OR1a 1.123 0.020 56.674 0.000 1.123 3.411
.T2_OD1a 1.196 0.024 49.981 0.000 1.196 3.009
.T2_OR1a 1.069 0.015 70.008 0.000 1.069 4.214
.V2_OD1a 1.319 0.028 46.930 0.000 1.319 2.825
.V2_OR1a 1.080 0.016 66.108 0.000 1.080 3.979
FA 0.000 0.000 0.000
Variances:
Estimate Std.Err z-value P(>|z|) Std.lv Std.all
.T1_OD1a 0.115 0.014 8.284 0.000 0.115 0.786
.T1_OR1a 0.133 0.015 9.007 0.000 0.133 0.919
.V1_OD1a 0.159 0.015 10.791 0.000 0.159 0.786
.V1_OR1a 0.089 0.013 6.742 0.000 0.089 0.823
.T2_OD1a 0.089 0.013 7.114 0.000 0.089 0.565
.T2_OR1a 0.056 0.011 5.086 0.000 0.056 0.866
.V2_OD1a 0.144 0.015 9.510 0.000 0.144 0.660
.V2_OR1a 0.061 0.011 5.468 0.000 0.061 0.832
FA 0.031 0.011 2.977 0.003 1.000 1.000
R-Square:
Estimate
T1_OD1a 0.214
T1_OR1a 0.081
V1_OD1a 0.214
V1_OR1a 0.177
T2_OD1a 0.435
T2_OR1a 0.134
V2_OD1a 0.340
V2_OR1a 0.168
Terrence Jorgensen
Sep 25, 2018, 7:06:39 AM9/25/18
to lavaan
Results using "NA*"> myModel11 <- "FA =~ NA*T1_OD1a + T1_OR1a + V1_OD1a + V1_OR1a + T2_OD1a + T2_OR1a + V2_OD1a + V2_OR1a" > fit11 <- cfa(myModel11, data=data.T1, estimator = "WLSMV", ordered = c("T1_OD1a, T1_OR1a, V1_OD1a, V1_OR1a, T2_OD1a, T2_OR1a, V2_OD1a, V2_OR1a"))
Set std.lv = TRUE if you want to free the first factor loading (then NA* would not be necessary anyway), or fix the factor variance in the model syntax. Either way, you need to identify the latent scale somehow, otherwise your model is not identified (perhaps causing the failure to calculate the robust test statistic, which prevents the remainder of the summary() from being printed).
Reply all
Reply to author
Forward
0 new messages