mixed models in GEE

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

mixed models in GEE

Kornbrot, Diana
hi
I wish to to ordinal regression with 1 between and 2 repeated factors, see syntax below
question 1. why cannot I not have either descriptives or means as output?
having to save the cumulative probabilities and get their means is a completely unnecessary hassle
question2. why are the model effects of different form the effects for their equivalent parameter?
which best represents the empirical results 

Really appreciate views oo ALL SPPS experts on this

best
Diana
________________________________________
Professor Diana Kornbrot
Work
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
+44 (0) 170 728 4626
[hidden email]
http://dianakornbrot.wordpress.com/
 http://go.herts.ac.uk/Diana_Kornbrot
skype:  kornbrotme
Home
19 Elmhurst Avenue
London N2 0LT, UK
 +44 (0) 208 444 2081                                                   



===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: mixed models in GEE

Maguin, Eugene

Re: question 1. It seems to me that this is something to take up with IBM-SPSS. I think I remember somebody from spss saying the genlinmixed was ‘copied’ from another spss program.

 

Question 2: (that sentence seems to have some words missing or mis-written). How about illustrating this with the syntax and some output?

Gene Maguin

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Kornbrot, Diana
Sent: Monday, March 20, 2017 8:08 AM
To: [hidden email]
Subject: mixed models in GEE

 

hi

I wish to to ordinal regression with 1 between and 2 repeated factors, see syntax below

question 1. why cannot I not have either descriptives or means as output?

having to save the cumulative probabilities and get their means is a completely unnecessary hassle

question2. why are the model effects of different form the effects for their equivalent parameter?

which best represents the empirical results 

 

Really appreciate views oo ALL SPPS experts on this

 

best

Diana

________________________________________
Professor Diana Kornbrot
Work
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
+44 (0) 170 728 4626
[hidden email]
http://dianakornbrot.wordpress.com/
 http://go.herts.ac.uk/Diana_Kornbrot
skype:  kornbrotme
Home
19 Elmhurst Avenue
London N2 0LT, UK
 +44 (0) 208 444 2081                                                   

 

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: mixed models in GEE

Kornbrot, Diana
Thanks Eugene
Q1, yes I was hoping SPSS would pick this up
Q2, script and output below

best
Diana
Script
* Generalized Estimating Equations.
GENLIN response9 (ORDER=ASCENDING) BY between3 rating select (ORDER=ASCENDING)
  /MODEL between3 rating select between3*rating between3*select rating*select
    between3*rating*select
 DISTRIBUTION=MULTINOMIAL LINK=CUMLOGIT
  /CRITERIA METHOD=FISHER(1) SCALE=1 MAXITERATIONS=100 MAXSTEPHALVING=5 PCONVERGE=1E-006(ABSOLUTE)
    SINGULAR=1E-012 ANALYSISTYPE=3(WALD) CILEVEL=95 LIKELIHOOD=FULL
  /REPEATED SUBJECT=Participant WITHINSUBJECT=rating*select SORT=YES CORRTYPE=UNSTRUCTURED
    ADJUSTCORR=YES COVB=MODEL MAXITERATIONS=100 PCONVERGE=1e-006(ABSOLUTE) UPDATECORR=1
  /MISSING CLASSMISSING=EXCLUDE
  /PRINT DESCRIPTIVES MODELINFO FIT SUMMARY SOLUTION
  /SAVE MEANPRED(:25) CIMEANPREDL(:25) CIMEANPREDU(:25).
Critical Output  want to know why Model Effects are so much larger than Parameter Effects although
    

Tests of Model Effects
Source
Type III
Wald Chi-Square
df
Sig.
Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented
1.740
2
.419
rating
7.663
1
.006
chosen,rejected
43.734
1
.000
Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented * rating
10.492
2
.005
Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented * chosen,rejected
19.517
2
.000
rating * chosen,rejected
6.664
1
.010
Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented * rating * chosen,rejected
2.451
2
.294
Dependent Variable: response9
Model: (Threshold), Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented, rating, chosen,rejected, Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented * rating, Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented * chosen,rejected, rating * chosen,rejected, Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented * rating 

Parameter Estimates
Parameter
B
Std. Error
95% Wald Confidence Interval
Hypothesis Test
Lower
Upper
Wald Chi-Square
df
Sig.
Threshold
[response9=1]
-3.287
.4856
-4.239
-2.335
45.825
1
.000
[response9=2]
-2.256
.3385
-2.919
-1.592
44.420
1
.000
[response9=3]
-1.652
.3082
-2.256
-1.048
28.729
1
.000
[response9=4]
-1.340
.3182
-1.963
-.716
17.726
1
.000
[response9=5]
-.202
.3182
-.825
.422
.402
1
.526
[response9=6]
1.154
.3127
.541
1.767
13.619
1
.000
[response9=7]
2.733
.3341
2.078
3.388
66.921
1
.000
[response9=8]
5.818
2.0595
1.781
9.854
7.979
1
.005
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=1]
.342
.3979
-.437
1.122
.741
1
.389
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=2]
.110
.4234
-.720
.940
.067
1
.795
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=3]
0a
.
.
.
.
.
.
[rating=1]
.761
.3890
-.002
1.523
3.824
1
.051
[rating=2]
0a
.
.
.
.
.
.
[chosen,rejected=1]
2.601
.4336
1.752
3.451
35.990
1
.000
[chosen,rejected=2]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=1] * [rating=1]
-.163
.5114
-1.165
.839
.102
1
.750
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=1] * [rating=2]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=2] * [rating=1]
.399
.5573
-.693
1.491
.512
1
.474
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=2] * [rating=2]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=3] * [rating=1]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=3] * [rating=2]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=1] * [chosen,rejected=1]
-1.738
.5224
-2.762
-.714
11.068
1
.001
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=1] * [chosen,rejected=2]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=2] * [chosen,rejected=1]
-1.903
.5544
-2.989
-.816
11.781
1
.001
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=2] * [chosen,rejected=2]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=3] * [chosen,rejected=1]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=3] * [chosen,rejected=2]
0a
.
.
.
.
.
.
[rating=1] * [chosen,rejected=1]
-1.483
.5615
-2.584
-.382
6.975
1
.008
[rating=1] * [chosen,rejected=2]
0a
.
.
.
.
.
.
[rating=2] * [chosen,rejected=1]
0a
.
.
.
.
.
.
[rating=2] * [chosen,rejected=2]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=1] * [rating=1] * [chosen,rejected=1]
1.050
.7173
-.356
2.456
2.141
1
.143
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=1] * [rating=1] * [chosen,rejected=2]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=1] * [rating=2] * [chosen,rejected=1]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=1] * [rating=2] * [chosen,rejected=2]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=2] * [rating=1] * [chosen,rejected=1]
.988
.7722
-.525
2.502
1.638
1
.201
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=2] * [rating=1] * [chosen,rejected=2]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=2] * [rating=2] * [chosen,rejected=1]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=2] * [rating=2] * [chosen,rejected=2]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=3] * [rating=1] * [chosen,rejected=1]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=3] * [rating=1] * [chosen,rejected=2]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=3] * [rating=2] * [chosen,rejected=1]
0a
.
.
.
.
.
.
[Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented=3] * [rating=2] * [chosen,rejected=2]
0a
.
.
.
.
.
.
(Scale)
1






Dependent Variable: response9
Model: (Threshold), Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented, rating, chosen,rejected, Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented * rating, Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented * chosen,rejected, rating * chosen,rejected, Condition: 1 Neutral; 2 Positive Non-Action; 3 Action Oriented * rating * chosen,rejected
a. Set to zero because this parameter is redundant.

* chosen,rejected

On 20 Mar 2017, at 16:08, Maguin, Eugene <[hidden email]> wrote:

Re: question 1. It seems to me that this is something to take up with IBM-SPSS. I think I remember somebody from spss saying the genlinmixed was ‘copied’ from another spss program. 
 
Question 2: (that sentence seems to have some words missing or mis-written). How about illustrating this with the syntax and some output?
Gene Maguin
 
From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Kornbrot, Diana
Sent: Monday, March 20, 2017 8:08 AM
To: [hidden email]
Subject: mixed models in GEE
 
hi 
I wish to to ordinal regression with 1 between and 2 repeated factors, see syntax below
question 1. why cannot I not have either descriptives or means as output?
having to save the cumulative probabilities and get their means is a completely unnecessary hassle
question2. why are the model effects of different form the effects for their equivalent parameter?
which best represents the empirical results 
 
Really appreciate views oo ALL SPPS experts on this
 
best
Diana

________________________________________
Professor Diana Kornbrot
Work
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
+44 (0) 170 728 4626
[hidden email]
http://dianakornbrot.wordpress.com/
 http://go.herts.ac.uk/Diana_Kornbrot
skype:  kornbrotme
Home
19 Elmhurst Avenue
London N2 0LT, UK
 +44 (0) 208 444 2081                                                   

 
===================== To manage your subscription to SPSSX-L, send a message to[hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD
===================== To manage your subscription to SPSSX-L, send a message to[hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

________________________________________
Professor Diana Kornbrot
Work
University of Hertfordshire
College Lane, Hatfield, Hertfordshire AL10 9AB, UK
+44 (0) 170 728 4626
[hidden email]
http://dianakornbrot.wordpress.com/
 http://go.herts.ac.uk/Diana_Kornbrot
skype:  kornbrotme
Home
19 Elmhurst Avenue
London N2 0LT, UK
 +44 (0) 208 444 2081                                                   



===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD
Loading...