Zero Variance Error from a Covariance... PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Steven Siciliano posted on Friday, April 13, 2012 - 7:40 pm
Hello,

I'm using a covariance matrix as an input for a SEM (see model below) in MPlus. It is commonly (but not always) returning an error code of zero variance in some variances for a covariance matrix in which the variance is not 0. Any ideas on why this is occurring?

VARIABLE:
NAMES ARE Moisture N_Fix BareGnd Bryoph Lich Forbs Gram Shrubs;
USEVARIABLES ARE Moisture N_Fix BareGnd Bryoph Lich Forbs Gram Shrubs;

ANALYSIS:
TYPE IS GENERAL;
ITERATIONS = 1000;
CONVERGENCE = 0.00005;

MODEL:
Forbs ON Moisture;
Shrubs ON Moisture;
Gram ON Moisture;
Bryoph ON Shrubs Gram Forbs Moisture ;
Lich ON Shrubs Gram Forbs Moisture Bryoph ;
BareGnd ON Shrubs Gram Forbs Lich Bryoph ;
N_Fix ON Bryoph Lich BareGnd;

Forbs WITH Shrubs;
Forbs WITH Gram;
Gram WITH Shrubs;

OUTPUT: STANDARDIZED MODINDICES(3.84);

*** ERROR
One or more variables have a variance of zero.
Variable Variance
**MOISTURE -0.010
N_FIX 0.000
BAREGND 0.001
BRYOPH 0.001
**LICH 0.000
FORBS 0.000
**GRAM 0.000
**SHRUBS -0.001
 Linda K. Muthen posted on Friday, April 13, 2012 - 8:15 pm
You don't show the DATA command but I suspect you are not using the TYPE option to indicate that the data are in the form of a covariance matrix. See Example 13.1 and the DATA command.
 Steven Siciliano posted on Friday, April 13, 2012 - 11:35 pm
Sorry about not posting the Data commands. Here they are. I am using the following Data Commands. Are these sufficient? I am using a full covariance matrix.

TITLE: SEM of Alex Fiord Lowland N-Fixation Data

DATA:
FILE IS "bin_1_491_obs.txt";
TYPE IS FULLCOV;
NOBSERVATIONS = 491;

VARIABLE:
NAMES ARE Moisture N_Fix BareGnd Bryoph Lich Forbs Gram Shrubs;
USEVARIABLES ARE Moisture N_Fix BareGnd Bryoph Lich Forbs Gram Shrubs;

EXAMPLE of Covariance for Moisture and N_Fix

Moisture N_Fix
Moisture -0.000929777 0.008661346
N_Fix 0.008661346 -0.010370599

ERROR MESSAGE
*** ERROR
One or more variables have a variance of zero.

Variable Variance

**MOISTURE -0.010
N_FIX 0.000
 Linda K. Muthen posted on Saturday, April 14, 2012 - 12:37 am
Please send the data, output, and your license number to support@statmodel.com.
 Joe posted on Monday, April 30, 2012 - 2:48 am
Hi,

I am trying to estimate a growth model so that the variance of the intercept is zero (I have already done i@0). In this case I am trying the following:

DEFINE:
t0=prfsb-prfsb;
t1=prfpm3s-prfsb;
t2=prfpm4s-prfsb;
t3=prfsm-prfsb;
t4=prfpm7s-prfsb;
t5=prfpm8s-prfsb;
t6=prfpm9s-prfsb;
t8=prfse-prfsb;

MODEL:
i s | t0@0 t1@1 t2@2 t3@3 t4@4 t5@5 t6@6 t8@8;

so that the score of each occasion has the first score subtracted from it. However I keep getting the following error message.

*** ERROR
One or more variables have a variance of zero. Check your data and format statement.

VARIANCE = NOCHECK did not work either because the "ESTIMATED COVARIANCE MATRIX was NON-INVERTIBLE."

Thank you for your suggestions.

Best,
Joe
 Linda K. Muthen posted on Monday, April 30, 2012 - 2:40 pm
Please send the output and your license number to support@statmodel.com.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: