problems in running LISREL programs

[researcher...@gmail.com]

Tis message is long but not complex-for someone who has used LISREL.

Subject-(1)Why do I get these error messages?
(2)Why do these programs run sometimes and not run at
other times?
I am trying to find out the discriminant validity between some
constructs.First, I run a programin which the correlation between the
constructs is unconstrained .Then,I run another program in which the
correlation between the constructs is fixed at one.
I am reproducing the 2 programs and the messages I get.Please note that
these programs have given me results at least twice. I do not know why
they gave me results and why mostly they do give me error messages.
Situation A) where correlation was unconstrained.
The program syntax is given below.

"TI

!DA NI=6 NO=101 NG=1 MA=CM
SY='C:\My Documents\abdisvalis\jpaffuncons\japecptaff2.dsf' NG=1
SE
1 2 3 4 5 6 /
MO NX=6 NK=2 LX=FU,FI PH=SY,FR TD=DI,FR
LK
aff1 aff2
FR LX(1,1) LX(2,1) LX(3,2) LX(4,2) LX(5,2) LX(6,2)
PD
OU ME=ML RS SS SC "

The message I get in the out file is
"The following lines were read from file C:\My
Documents\abdisvalis\jpaffuncons\try1.LPJ:
TI
!DA NI=6 NO=101 NG=1 MA=CM
SY='C:\My Documents\abdisvalis\jpaffuncons\japecptaff2.dsf' NG=1"

When I remove the exclamation mark before 'DA'(The manual does not
mention any need for an exclamation mark .)I get different error
messages.

"model does not converge"

"W_A_R_N_I_N_G: Matrix to be analyzed is not positive definite,
ridge option taken with ridge constant = 10.000

TI

Covariance Matrix
Note: This matrix is diagonal.

VAR 1 VAR 2 VAR 3 VAR 4 VAR 5
VAR 6
-------- -------- -------- -------- --------
--------
11.00 - - - - - - - -
- -

F_A_T_A_L E_R_R_O_R: Matrix to be analyzed is not positive
definite."

Situation B.)where correlation between constructs is set at 1.0
The program is as follows:-
"TI

!DA NI=6 NO=101 NG=1 MA=CM
SY='C:\My Documents\abdisvalis\jpaffuncons\japecptaff2.dsf' NG=1
SE
1 2 3 4 5 6 /
MO NX=6 NK=2 LX=FU,FI PH=SY,FR TD=DI,FR
LK
aff1 aff2
FR LX(1,1) LX(2,1) LX(3,2) LX(4,2) LX(5,2) LX(6,2)
VA 1.0 PH(2,1)
FI PH(2,1)
PD
OU ME=ML RS SS SC "
The error messages I get are the same as those for the 2 versions of
the first program.
I do not know how to solve this problem. Please help me.

[neila...@msn.com]

SPSS hasn't had LISREL as a subprogram for many years!!!
Maybe blow this by the folks at Scientific Software??
-But it looks like a specification issue (I haven't used LISREL in
AGES!!!)
Google LISREL doesn't yield too many penetrating insights.
>From a quick peek your model does not appear to be properly identified!
Correlation of 1??? Of course!! Think on that abit! (Maybe PH 1 1 and
PH 2 2 instead?)
Neila