Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: sem
From
"Airey, David C" <[email protected]>
To
"[email protected]" <[email protected]>
Subject
Re: st: sem
Date
Sat, 2 Mar 2013 03:35:08 +0000
.
Following up of Jeff's and Billy's replies
http://www.stata.com/statalist/archive/2013-02/msg01056.html
I need to investigate this further, because indeed using the
satopts(tech(dfp))
option instead of assuming the saturated model won't converge with
satopts(iter(10))
will allow convergence and all the post-estimation stats
estat gof, stats(all)
There are some differences in the baseline and saturated
models fit statistics in Stata and Mplus (below), particulary
the df.
Also, I found this useful (for me) page on the saturated and baseline
models in Stata:
http://www.ats.ucla.edu/stat/stata/faq/sem_baseline.htm
-Dave
. sem (read6 <- Intercept@1 Slope@0 _cons@0) ///
> (read7 <- Intercept@1 Slope@1 _cons@0) ///
> (read8 <- Intercept@1 Slope@2 _cons@0) ///
> (read9 <- Intercept@1 Slope@3 _cons@0) ///
> (read10 <- Intercept@1 Slope@4 _cons@0) ///
> (read11 <- Intercept@1 Slope@5 _cons@0) ///
> (read12 <- Intercept@1 Slope@6 _cons@0) ///
> (read13 <- Intercept@1 Slope@7 _cons@0) ///
> (read14 <- Intercept@1 Slope@8 _cons@0), ///
> latent(Intercept Slope) ///
> cov(Intercept*Slope) ///
> var(e.read6@var e.read7@var e.read8@var e.read9@var ///
> e.read10@var e.read11@var e.read12@var e.read13@var ///
> e.read14@var) ///
> means(Intercept Slope) ///
> method(mlmv) ///
> satopts(tech(dfp))
-------------------------------------------------------------------------------
| OIM
| Coef. Std. Err. z P>|z| [95% Conf. Interval]
--------------+----------------------------------------------------------------
Measurement |
read6 <- |
Intercept | 1 (constrained)
_cons | 0 (constrained)
------------+----------------------------------------------------------------
read7 <- |
Intercept | 1 (constrained)
Slope | 1 (constrained)
_cons | 0 (constrained)
------------+----------------------------------------------------------------
read8 <- |
Intercept | 1 (constrained)
Slope | 2 (constrained)
_cons | 0 (constrained)
------------+----------------------------------------------------------------
read9 <- |
Intercept | 1 (constrained)
Slope | 3 (constrained)
_cons | 0 (constrained)
------------+----------------------------------------------------------------
read10 <- |
Intercept | 1 (constrained)
Slope | 4 (constrained)
_cons | 0 (constrained)
------------+----------------------------------------------------------------
read11 <- |
Intercept | 1 (constrained)
Slope | 5 (constrained)
_cons | 0 (constrained)
------------+----------------------------------------------------------------
read12 <- |
Intercept | 1 (constrained)
Slope | 6 (constrained)
_cons | 0 (constrained)
------------+----------------------------------------------------------------
read13 <- |
Intercept | 1 (constrained)
Slope | 7 (constrained)
_cons | 0 (constrained)
------------+----------------------------------------------------------------
read14 <- |
Intercept | 1 (constrained)
Slope | 8 (constrained)
_cons | 0 (constrained)
--------------+----------------------------------------------------------------
Mean |
Intercept | 21.83271 .5694494 38.34 0.000 20.71661 22.94881
Slope | 5.13628 .1159911 44.28 0.000 4.908942 5.363619
--------------+----------------------------------------------------------------
Variance |
e.read6 | 28.2614 1.944146 24.69666 32.34068
e.read7 | 28.2614 1.944146 24.69666 32.34068
e.read8 | 28.2614 1.944146 24.69666 32.34068
e.read9 | 28.2614 1.944146 24.69666 32.34068
e.read10 | 28.2614 1.944146 24.69666 32.34068
e.read11 | 28.2614 1.944146 24.69666 32.34068
e.read12 | 28.2614 1.944146 24.69666 32.34068
e.read13 | 28.2614 1.944146 24.69666 32.34068
e.read14 | 28.2614 1.944146 24.69666 32.34068
Intercept | 40.17873 7.117793 28.39247 56.85771
Slope | 1.62987 .290664 1.149088 2.311813
--------------+----------------------------------------------------------------
Covariance |
Intercept |
Slope | .8008819 1.079894 0.74 0.458 -1.315672 2.917436
-------------------------------------------------------------------------------
LR test of model vs. saturated: chi2(48) = 142.26, Prob > chi2 = 0.0000
. estat gof, stats(all)
----------------------------------------------------------------------------
Fit statistic | Value Description
---------------------+------------------------------------------------------
Likelihood ratio |
chi2_ms(48) | 142.256 model vs. saturated
p > chi2 | 0.000
chi2_bs(36) | 602.769 baseline vs. saturated
p > chi2 | 0.000
---------------------+------------------------------------------------------
Population error |
RMSEA | 0.096 Root mean squared error of approximation
90% CI, lower bound | 0.078
upper bound | 0.114
pclose | 0.000 Probability RMSEA <= 0.05
---------------------+------------------------------------------------------
Information criteria |
AIC | 5946.830 Akaike's information criterion
BIC | 5967.026 Bayesian information criterion
---------------------+------------------------------------------------------
Baseline comparison |
CFI | 0.834 Comparative fit index
TLI | 0.875 Tucker-Lewis index
---------------------+------------------------------------------------------
Size of residuals |
CD | 0.986 Coefficient of determination
----------------------------------------------------------------------------
Note: SRMR is not reported because of missing values.
Mplus 7:
TITLE: Unconditional LGM of read data by age with random intercept,
random slope, equal level-1 error variance across time
DATA: file is byage.csv;
VARIABLE:
Names are id read6-read14;
Usevariables are read6-read14;
missing are all (-99);
ANALYSIS:
type=missing meanstructure;
estimator=ml;
coverage=0;
MODEL:
int by read6@1 read7@1 read8@1 read9@1
read10@1 read11@1 read12@1 read13@1 read14@1;
linear by read6@0 read7@1 read8@2 read9@3
read10@4 read11@5 read12@6 read13@7 read14@8;
read6-read14 (1);
[read6-read14@0];
int linear;
[int linear];
int with linear;
MODEL RESULTS
Two-Tailed
Estimate S.E. Est./S.E. P-Value
INT BY
READ6 1.000 0.000 999.000 999.000
READ7 1.000 0.000 999.000 999.000
READ8 1.000 0.000 999.000 999.000
READ9 1.000 0.000 999.000 999.000
READ10 1.000 0.000 999.000 999.000
READ11 1.000 0.000 999.000 999.000
READ12 1.000 0.000 999.000 999.000
READ13 1.000 0.000 999.000 999.000
READ14 1.000 0.000 999.000 999.000
LINEAR BY
READ6 0.000 0.000 999.000 999.000
READ7 1.000 0.000 999.000 999.000
READ8 2.000 0.000 999.000 999.000
READ9 3.000 0.000 999.000 999.000
READ10 4.000 0.000 999.000 999.000
READ11 5.000 0.000 999.000 999.000
READ12 6.000 0.000 999.000 999.000
READ13 7.000 0.000 999.000 999.000
READ14 8.000 0.000 999.000 999.000
INT WITH
LINEAR 0.801 1.080 0.742 0.458
Means
INT 21.833 0.569 38.340 0.000
LINEAR 5.136 0.116 44.282 0.000
Intercepts
READ6 0.000 0.000 999.000 999.000
READ7 0.000 0.000 999.000 999.000
READ8 0.000 0.000 999.000 999.000
READ9 0.000 0.000 999.000 999.000
READ10 0.000 0.000 999.000 999.000
READ11 0.000 0.000 999.000 999.000
READ12 0.000 0.000 999.000 999.000
READ13 0.000 0.000 999.000 999.000
READ14 0.000 0.000 999.000 999.000
Variances
INT 40.179 7.118 5.645 0.000
LINEAR 1.630 0.291 5.607 0.000
Residual Variances
READ6 28.261 1.944 14.537 0.000
READ7 28.261 1.944 14.537 0.000
READ8 28.261 1.944 14.537 0.000
READ9 28.261 1.944 14.537 0.000
READ10 28.261 1.944 14.537 0.000
READ11 28.261 1.944 14.537 0.000
READ12 28.261 1.944 14.537 0.000
READ13 28.261 1.944 14.537 0.000
READ14 28.261 1.944 14.537 0.000
OUTPUT:
TECH1;
MODEL FIT INFORMATION
Number of Free Parameters 6
Loglikelihood
H0 Value -2967.415
H1 Value -2896.703
Information Criteria
Akaike (AIC) 5946.830
Bayesian (BIC) 5967.026
Sample-Size Adjusted BIC 5948.013
(n* = (n + 2) / 24)
Chi-Square Test of Model Fit
Value 141.423
Degrees of Freedom 44
P-Value 0.0000
RMSEA (Root Mean Square Error Of Approximation)
Estimate 0.102
90 Percent C.I. 0.083 0.121
Probability RMSEA <= .05 0.000
CFI/TLI
CFI 0.829
TLI 0.876
Chi-Square Test of Model Fit for the Baseline Model
Value 601.936
Degrees of Freedom 32
P-Value 0.0000
SRMR (Standardized Root Mean Square Residual)
Value
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/faqs/resources/statalist-faq/
* http://www.ats.ucla.edu/stat/stata/