Nominal unordered variable and SEM
Message/Author
 YILDIZ HELENE posted on Friday, January 03, 2014 - 1:11 pm
Dear Professor,

We are working on a project where the aim is to find the consumer behaviour.
This variable takes 4 behaviour types unrelated.
so my nominal variable has four unordered categories

this variable is impacted by :
- individuals characteristics (Male/Female; age; formation)
-motivations (represented by 4 first-order factors)

so the model takes this equation =

internet behaviour/magazin behaviour/retail offline behaviour/retail online behaviour = motivation1(item11, item12,item13) + motivation2(item21, item22, item23)+...+ age (5categories) + Sex (male, femal)...

can I use a SEM with a nominal variable has four unordered categories ?

I have looked through the manual, but have not seen any examples.
If you suggest appropriate method ...
or an axample to program this model..

Thank you and happy new year

Respectfully yours,

I thank you for your response,

Helene
 Linda K. Muthen posted on Friday, January 03, 2014 - 2:35 pm
Use Example 3.6. Add factors as shown in Example 5.1. The factors and other covariates can be used on the right-hand side of the ON statement in the multinomial logistic regression.
 YILDIZ HELENE posted on Saturday, January 04, 2014 - 9:51 am
I thank you for your quickly message !
 YILDIZ HELENE posted on Sunday, January 05, 2014 - 2:24 pm
Dear Professor,

I did the model (Vary is my nominal variable with 8 categories) but I have this message :

TITLE: a
multinomial
logistic regression
for an
unordered
categorical (nominal)
dependent variable
with 11 covariate

DATA:
FILE IS Alldata12.dat;

VARIABLE:

NAMES ARE AGE SEX PROF BP1-BP2
BI1-BI2 AR1-AR3 CP1-CP4 BD1-BD3
BIS1-BIS3 BER1-BER3 PCAT VARY VARX;

USEVARIABLES ARE AGE SEX PROF
BP1-BP2 BI1-BI2 AR1-AR3 CP1-CP4
BD1-BD3 BIS1-BIS3 BER1-BER3
PCAT VARY;
NOMINAL IS VARY;

ANALYSIS:

ESTIMATOR = MLR;

MODEL:

BI BY BI1-BI2;
BP BY BP1-BP2;
AR BY AR1-AR3;
CP BY CP1-CP4;
BD BY BD1-BD3;
BIS BY BIS1-BIS3;
BER BY BER1-BER3;
VARY#1 VARY#2 VARY#3
VARY#4 VARY#5 VARY#6
VARY#7 VARY#8 ON AGE
SEX PROF BP AR CP
BD BIS BER PCAT;

OUTPUT:

MODINDICES(7);
TECH4;

*** ERROR in MODEL command
Unknown threshold for NOMINAL variable VARY: VARY#8
 YILDIZ HELENE posted on Sunday, January 05, 2014 - 2:50 pm
AGE SEX PROF BP1-BP2
BI1-BI2 AR1-AR3 CP1-CP4 BD1-BD3
BIS1-BIS3 BER1-BER3 PCAT VARY

where:
prof : is the level of respondent, so it's a categorical variable
PCAT : is the product type (there are 6 categories)

 Linda K. Muthen posted on Sunday, January 05, 2014 - 3:14 pm
How many categories does vary have. The last category is the reference category and should not be referred to.

Posts longer than one window are not appropriate for Mplus Discussion.
 YILDIZ HELENE posted on Sunday, January 05, 2014 - 11:29 pm
my nominal variable has 8 categories
when I take of one of them I have this error message
 YILDIZ HELENE posted on Sunday, January 05, 2014 - 11:29 pm
MODEL:

BI BY BI1-BI2;
BP BY BP1-BP2;
AR BY AR1-AR3;
CP BY CP1-CP4;
BD BY BD1-BD3;
BIS BY BIS1-BIS3;
BER BY BER1-BER3;
VARY#1 VARY#2 VARY#3
VARY#4 VARY#5 VARY#6
VARY#7 ON AGE
SEX PROF BP AR CP
BD BIS BER PCAT;
OUTPUT:
MODINDICES(7);
TECH4;

*** ERROR in MODEL command
Variances for nominal variables are not currently defined.
Variance given for: VARY#1
*** ERROR in MODEL command
Variances for nominal variables are not currently defined.
Variance given for: VARY#2
*** ERROR in MODEL command
Variances for nominal variables are not currently defined.
Variance given for: VARY#3
*** ERROR in MODEL command
Variances for nominal variables are not currently defined.
Variance given for: VARY#4
*** ERROR in MODEL command
Variances for nominal variables are not currently defined.
Variance given for: VARY#5
*** ERROR in MODEL command
Variances for nominal variables are not currently defined.
Variance given for: VARY#6
 Linda K. Muthen posted on Monday, January 06, 2014 - 7:03 am
Variances are not model parameters for nominal variables. Remove the variance statements.