LISREL syntax error

[ecnusbau]

Hi all,

I'm looking for help with my syntax testing invariance of means and intercepts across two groups. The data was teacher-supplied, so my problem is DEFINITELY somewhere in my syntax, which is posted below. I keep getting errors saying my phi matrix is not positive definite, and I have no idea how to fix it. Any ideas?

In each group, there are 2 latent variables with 4 observed x variables each.

TITLE=READING AND ENGLISH SELF-EFFICACY
DA NO=250 NI=8 MA=CM NG=2
CM
93.3686022692858
79.5338745541415 79.9300617511353
84.5694040780116 77.6291446914262 88.1355015855679
87.1050232547505 78.7015858433717 84.6265355563024 91.4808027173393
9.3781664401782 9.37139963322129 10.5294931610047 10.6279925352157 114.321180096987
7.60371289048078 7.87348101198644 8.26848210063562 9.0997727853419 103.430749337207 102.422527720617
7.94626429392916 6.21407915639345 7.84639030229656 7.26713716473023 93.1944011803652 C
86.5944396659961 84.5905721790042
10.3032813261378 9.36586459609417 10.5808667591251 11.2039113378897 87.8516616367962 C
82.4978303787938 74.6862740950884 79.8884976722131
ME
10.1346968054826 11.8949126191111 11.0183232499892 10.5456440115764 10.3739591221876 12.4896556526254 11.4887874944207 10.6851894464758
LA
ENGSE1 ENGSE2 ENGSE3 ENGSE4 READ1 READ2 READ3 READ4
MO NX=8 NK=2 KA=FI LX=FU,FI TX=FR PH=SY,FR
VA 0 KA 1 KA 2
VA 1 LX 1 1
FR LX 2 1 LX 3 1 LX 4 1
VA 1 LX 1 2
FR LX 2 2 LX 3 2 LX 4 2
LK
ENG_SE READ_INT
PD
OU MI
DA NO=250 NI=8 MA=CM NG=2
CM
57.0224984356011
41.1901493366287 38.4978924011485
57.1801454625205 44.9899816026644 66.6982708129778
56.9357402383736 43.9811055944875 62.4650126903283 64.8576252799913
10.7938184477448 10.2703508846233 14.0957161031358 13.2529293568461 61.3623127839341
12.7263139503040 11.3491136172275 16.1934614116803 14.9641095297885 61.3464924701357 71.0973283500895
12.8387933851309 12.3189431929358 17.8922328148724 16.7946051349406 62.6009544458689 C
68.9771822687551 74.1067260217183
11.8690743603170 11.4726068377400 15.7182317804730 14.6671339936057 63.9320636871307 C
69.8779505294744 71.1841110344282 75.0422890780057
ME
14.0349167273324 16.0043752935715 15.0718154085047 14.7120105357764 13.6530270557572 15.5947082357896 14.3773009360619 13.7912065914032
LA
ENGSE1 ENGSE2 ENGSE3 ENGSE4 READ1 READ2 READ3 READ4
MO NX=8 NK=2 PH=PS LX=IN TX=IN KA=FR
PD
OU MI

 

[wandatheavenger]

You made a mistake in setting scale and freeing your paths for the second latent factor (Read_Int) by having it share observed variables with English Self Efficacy. Your first set of VA and FR statements are correct, but the second set should start with VA 1 LX 5 2 and move on into FR LX 6 2.... When you have problems like this, an easy way to diagnose them is to add ME=TSLS in your OU line. This has the model estimate using two-stage least squares, which almost always at least gets it to draw a path diagram. This way, you can visually check your model to ensure that it is specified correctly prior to actually estimating it with maximum likelihood (to get back to MLE, just delete the ME=TSLS part once everything's properly specified).

 

[ecnusbau]

it always seems to be little things like that that get me... thanks so much for lending me a second set of eyes!