PLUM - Ordinal Regression

Notes

Output Created 10-JAN-2017 15:59:45

Comments
D:\IAN\Kuliah\Semester
Data
V\ADK\adk fix.sav
Active Dataset DataSet1
Input Filter <none>
Weight count
Split File <none>
N of Rows in Working Data File 16
User-defined missing values are
Definition of Missing
treated as missing.
Missing Value Handling Statistics are based on all cases
Cases Used with valid data for all variables in
the model.
PLUM inf BY gdp un
/CRITERIA=CIN(95) DELTA(0)
LCONVERGE(0) MXITER(100)
MXSTEP(5) PCONVERGE(1.0E-
Syntax 6) SINGULAR(1.0E-8)
/LINK=LOGIT
/PRINT=CELLINFO CORB
COVB FIT PARAMETER
SUMMARY TPARALLEL.
Processor Time 00:00:00,22
Resources
Elapsed Time 00:00:00,33

[DataSet1] D:\IAN\Kuliah\Semester V\ADK\adk fix.sav

Warnings

There are 8 (33,3%) cells (i.e., dependent variable levels by combinations of

predictor variable values) with zero frequencies.
Unexpected singularities in the Fisher Information matrix are encountered.
There may be a quasi-complete separation in the data. Some parameter
estimates will tend to infinity.
The PLUM procedure continues despite the above warning(s). Subsequent
results shown are based on the last iteration. Validity of the model fit is
uncertain.

Case Processing Summary

N Marginal
Percentage

low 152 91,6%

inf middle 11 6,6%

high 3 1,8%
low 27 16,3%
low middle 42 25,3%
gdp
upper middle 46 27,7%
high 51 30,7%
low 66 39,8%
un
high 100 60,2%
Valid 166 100,0%
Missing 0
Total 166

Model Fitting Information

Model -2 Log Likelihood Chi-Square df Sig.

Intercept Only 35,610

Final 23,164 12,446 4 ,014

Link function: Logit.

Goodness-of-Fit

Chi-Square df Sig.

Pearson 3,522 10 ,966

Deviance 4,465 10 ,924

Link function: Logit.

Pseudo R-Square
Cox and Snell ,072
Nagelkerke ,149
McFadden ,113

Link function: Logit.

Parameter Estimates

Estimate Std. Error Wald df Sig. 95% Confidence Interval

Lower Bound Upper Bound

[inf = 1,0] 21,272 ,484 1934,343 1 ,000 20,324 22,220

Threshold
[inf = 2,0] 22,929 ,703 1062,811 1 ,000 21,550 24,307
[gdp=1,00] 19,460 ,786 612,620 1 ,000 17,919 21,001

[gdp=2,00] 19,789 ,658 905,585 1 ,000 18,500 21,078

[gdp=3,00] 19,362 ,000 . 1 . 19,362 19,362

Location
[gdp=4,00] 0a . . 0 . . .

[un=1,00] -,717 ,640 1,255 1 ,263 -1,972 ,538

[un=2,00] 0a . . 0 . . .

Link function: Logit.

a. This parameter is set to zero because it is redundant.

Asymptotic Correlation Matrix

Threshold Location

[inf = 1,0] [inf = 2,0] [gdp=1,00] [gdp=2,00] [gdp=3,00] [gdp=4,00] [un=1,00] [un=2,00]

[inf = 1,0] 1,000 ,662 ,548 ,648 . .a ,239 .a

Threshold
[inf = 2,0] ,662 1,000 ,378 ,455 . .a ,151 .a
[gdp=1,00] ,548 ,378 1,000 ,451 . .a -,133 .a

[gdp=2,00] ,648 ,455 ,451 1,000 . .a -,187 .a

[gdp=3,00] . . . . 1,000 .a . .a
Location
[gdp=4,00] .a .a .a .a .a .a .a .a

[un=1,00] ,239 ,151 -,133 -,187 . .a 1,000 .a

[un=2,00] .a .a .a .a .a .a .a .a

Link function: Logit.

a. One or both parameter estimates are redundant.

Asymptotic Covariance Matrix

Threshold Location
 [inf = 1,0] [inf = 2,0] [gdp=1,00] [gdp=2,00] [gdp=3,00] [gdp=4,00] [un=1,00] [un=2,00]
a
[inf = 1,0] ,234 ,225 ,208 ,206 ,000 0 ,074 0a
Threshold
[inf = 2,0] ,225 ,495 ,209 ,210 ,000 0a ,068 0a
[gdp=1,00] ,208 ,209 ,618 ,233 ,000 0a -,067 0a

[gdp=2,00] ,206 ,210 ,233 ,432 ,000 0a -,079 0a

[gdp=3,00] ,000 ,000 ,000 ,000 ,000 0a ,000 0a

Location
[gdp=4,00] 0a 0a 0a 0a 0a 0a 0a 0a

[un=1,00] ,074 ,068 -,067 -,079 ,000 0a ,410 0a

[un=2,00] 0a 0a 0a 0a 0a 0a 0a 0a

Link function: Logit.

a. One or both parameter estimates are redundant.

Cell Information
Frequency

gdp un inf

low middle high

Observed 12 1 0

low Expected 12,040 ,765 ,195

Pearson Residual -,043 ,277 -,445

low
Observed 12 2 0

high Expected 12,035 1,542 ,423

Pearson Residual -,027 ,391 -,660

Observed 19 2 0
low Expected 18,906 1,659 ,435
Pearson Residual ,068 ,276 -,666
low middle
Observed 17 3 1
high Expected 17,116 3,012 ,871
Pearson Residual -,065 -,008 ,141
Observed 12 1 0
low Expected 12,123 ,700 ,177
Pearson Residual -,136 ,369 -,423
upper middle
Observed 29 2 2
high Expected 28,742 3,351 ,907
Pearson Residual ,134 -,779 1,164
high Observed 19 0 0

low Expected 19,000 ,000 ,000

Pearson Residual ,000 ,000 ,000

high Observed 32 0 0

Expected 32,000 ,000 ,000

 Pearson Residual ,000 ,000 ,000

Link function: Logit.

Test of Parallel Linesa

Model -2 Log Likelihood Chi-Square df Sig.

Null Hypothesis 23,164

General ,000b 23,164 4 ,000

The null hypothesis states that the location parameters (slope coefficients) are the
same across response categories.
a. Link function: Logit.
b. The log-likelihood value is practically zero. There may be a complete separation in
the data. The maximum likelihood estimates do not exist.

Nominal Regression

Notes

Output Created 10-JAN-2017 16:02:41

Comments
D:\IAN\Kuliah\Semester
Data
V\ADK\adk fix.sav
Active Dataset DataSet1
Input Filter <none>
Weight count
Split File <none>
N of Rows in Working Data File 16
User-defined missing values are
Definition of Missing
treated as missing.
Missing Value Handling Statistics are based on all cases
Cases Used with valid data for all variables in
the model.
 NOMREG inf (BASE=FIRST
ORDER=ASCENDING) BY gdp
un
/CRITERIA CIN(95) DELTA(0)
MXITER(100) MXSTEP(5)
CHKSEP(20) LCONVERGE(0)
PCONVERGE(0.000001)
SINGULAR(0.00000001)
/MODEL
/STEPWISE=PIN(.05)
Syntax
POUT(0.1) MINEFFECT(0)
RULE(SINGLE)
ENTRYMETHOD(LR)
REMOVALMETHOD(LR)
/INTERCEPT=INCLUDE
/PRINT=ASSOCIATION
CELLPROB CLASSTABLE FIT
CORB COVB PARAMETER
SUMMARY LRT CPS STEP MFI
IC.
Processor Time 00:00:00,89
Resources
Elapsed Time 00:00:00,88

[DataSet1] D:\IAN\Kuliah\Semester V\ADK\adk fix.sav

Warnings

There are 8 (33,3%) cells (i.e., dependent variable levels by subpopulations)

with zero frequencies.
Unexpected singularities in the Hessian matrix are encountered. This
indicates that either some predictor variables should be excluded or some
categories should be merged.
The NOMREG procedure continues despite the above warning(s).
Subsequent results shown are based on the last iteration. Validity of the
model fit is uncertain.
Measures of Monotone Association table is not generated because the
dependent variable does not have exactly two levels.

Case Processing Summary

 N Marginal
Percentage

low 152 91,6%

inf middle 11 6,6%

high 3 1,8%
low 27 16,3%
low middle 42 25,3%
gdp
upper middle 46 27,7%
high 51 30,7%
low 66 39,8%
un
high 100 60,2%
Valid 166 100,0%
Missing 0
Total 166
Subpopulation 8a

a. The dependent variable has only one value observed in 2

(25,0%) subpopulations.

Model Fitting Information

Model Model Fitting Criteria Likelihood Ratio Tests

AIC BIC -2 Log Likelihood Chi-Square df Sig.

Intercept Only 39,610 45,834 35,610

Final 38,968 70,087 18,968 16,643 8 ,034

Goodness-of-Fit

Chi-Square df Sig.

Pearson ,285 6 1,000

Deviance ,268 6 1,000

Pseudo R-Square

Cox and Snell ,095

Nagelkerke ,196
McFadden ,151

Likelihood Ratio Tests

Effect Model Fitting Criteria Likelihood Ratio Tests

AIC of Reduced BIC of Reduced -2 Log Likelihood Chi-Square df Sig.

Model Model of Reduced Model

Intercept 38,968 70,087 18,968a ,000 0 .

gdp 40,449 52,897 32,449 13,481 6 ,036
un 38,087 62,983 22,087 3,119 2 ,210

The chi-square statistic is the difference in -2 log-likelihoods between the final model and a reduced model. The
reduced model is formed by omitting an effect from the final model. The null hypothesis is that all parameters of
that effect are 0.
a. This reduced model is equivalent to the final model because omitting the effect does not increase the degrees of
freedom.

Parameter Estimates

infa B Std. Error Wald df Sig. Exp(B) 95% Confidence Interval for Exp(B)

Lower Bound Upper Bound

Intercept -20,629 4502,895 ,000 1 ,996

[gdp=1,00] 18,722 4502,895 ,000 1 ,997 135195686,129 ,000 .b

[gdp=2,00] 18,840 4502,895 ,000 1 ,997 152078706,900 ,000 .b

middle [gdp=3,00] 18,116 4502,895 ,000 1 ,997 73716715,365 ,000 .b

[gdp=4,00] 0c . . 0 . . . .

[un=1,00] -,390 ,672 ,336 1 ,562 ,677 ,181 2,529

[un=2,00] 0c . . 0 . . . .
Intercept -21,321 ,731 850,656 1 ,000

[gdp=1,00] ,403 9983,878 ,000 1 1,000 1,497 ,000 .b

[gdp=2,00] 18,480 1,261 214,802 1 ,000 106122748,093 8964299,372 1256321012,504

high [gdp=3,00] 18,658 ,000 . 1 . 126826282,265 126826282,265 126826282,265

[gdp=4,00] 0c . . 0 . . . .
b
[un=1,00] -18,056 5951,728 ,000 1 ,998 1,440E-008 ,000 .

[un=2,00] 0c . . 0 . . . .

a. The reference category is: low.

b. Floating point overflow occurred while computing this statistic. Its value is therefore set to system missing.
c. This parameter is set to zero because it is redundant.

Asymptotic Correlation Matrix

infa inf

middle high

Intercept [gdp=1,00] [gdp=2,00] [gdp=3,00] [gdp=4,00] [un=1,00] [un=2,00] Intercept [gdp=1,00] [gdp=2,00] [gdp=3,00] [gdp=4,00] [
 Intercept 1 -1,000 -1,000 -1,000 .b ,000 .b ,000 ,000 ,000 . .b

[gdp=1,00] -1,000 1 1,000 1,000 .b ,000 .b ,000 ,000 ,000 . .b

[gdp=2,00] -1,000 1,000 1 1,000 .b ,000 .b ,000 ,000 ,000 . .b

middle [gdp=3,00] -1,000 1,000 1,000 1 .b ,000 .b ,000 ,000 ,000 . .b

[gdp=4,00] .b .b .b .b .b .b .b .b .b .b .b .b

[un=1,00] ,000 ,000 ,000 ,000 .b 1 .b -,016 ,000 -,024 . .b

[un=2,00] .b .b .b .b .b .b .b .b .b .b .b .b
Intercept ,000 ,000 ,000 ,000 .b -,016 .b 1 ,000 -,579 . .b

[gdp=1,00] ,000 ,000 ,000 ,000 .b ,000 .b ,000 1 ,000 . .b

[gdp=2,00] ,000 ,000 ,000 ,000 .b -,024 .b -,579 ,000 1 . .b

high [gdp=3,00] . . . . .b . .b . . . . .b

[gdp=4,00] .b .b .b .b .b .b .b .b .b .b .b .b

[un=1,00] ,000 ,000 ,000 ,000 .b ,000 .b ,000 ,000 ,000 . .b

[un=2,00] .b .b .b .b .b .b .b .b .b .b .b .b

a. The reference category is: low.

b. One or both parameter estimates are redundant.

Asymptotic Covariance Matrixa

infb inf

middle hig

Intercept [gdp=1,00] [gdp=2,00] [gdp=3,00] [gdp=4,00] [un=1,00] [un=2,00] Intercept [gdp=1,00] [gdp=2,00] [gdp=3
c c
Intercept 20276062,499 -20276062,446 -20276062,442 -20276062,469 0 -,130 0 ,002 ,005 ,006

[gdp=1,00] -20276062,446 20276062,844 20276062,470 20276062,458 0c -,054 0c ,001 ,051 ,002

c c
[gdp=2,00] -20276062,442 20276062,470 20276062,701 20276062,457 0 -,069 0 ,001 ,002 ,036

middle [gdp=3,00] -20276062,469 20276062,458 20276062,457 20276062,821 0c ,026 0c ,026 -,028 -,028
c c c c c c c c c
[gdp=4,00] 0 0 0 0 0 0 0 0 0 0c

[un=1,00] -,130 -,054 -,069 ,026 0c ,452 0c -,008 -,016 -,021

c c c c c c c c c
[un=2,00] 0 0 0 0 0 0 0 0 0 0c
Intercept ,002 ,001 ,001 ,026 0c -,008 0c ,534 -,534 -,534

[gdp=1,00] ,005 ,051 ,002 -,028 0c -,016 0c -,534 99677821,385 ,535

c c
[gdp=2,00] ,006 ,002 ,036 -,028 0 -,021 0 -,534 ,535 1,590

high [gdp=3,00] ,000 ,000 ,000 ,000 0c ,000 0c ,000 ,000 ,000
c c c c c c c c c
[gdp=4,00] 0 0 0 0 0 0 0 0 0 0c

[un=1,00] -,012 -,005 -,012 -,001 0c ,042 0c -,234 -,262 -,359

c c c c c c c c c
[un=2,00] 0 0 0 0 0 0 0 0 0 0c

a. There is no overdispersion adjustment.

b. The reference category is: low.
c. One or both parameter estimates are redundant.
 Classification

Observed Predicted

low middle high Percent Correct

low 152 0 0 100,0%

middle 11 0 0 0,0%
high 3 0 0 0,0%
Overall Percentage 100,0% 0,0% 0,0% 91,6%

Observed and Predicted Frequencies

un gdp inf Frequency Percentage

Observed Predicted Pearson Residual Observed Predicted

low 12 11,811 ,181 92,3% 90,9%

low middle 1 1,189 -,181 7,7% 9,1%

high 0 ,000 ,000 0,0% 0,0%

low 19 18,865 ,098 90,5% 89,8%

low middle middle 2 2,135 -,098 9,5% 10,2%

high 0 ,000 ,000 0,0% 0,0%

low
low 12 12,324 -,404 92,3% 94,8%

upper middle middle 1 ,676 ,404 7,7% 5,2%

high 0 ,000 ,000 0,0% 0,0%

low 19 19,000 ,000 100,0% 100,0%

high middle 0 ,000 ,000 0,0% 0,0%

high 0 ,000 ,000 0,0% 0,0%

low 12 12,189 -,150 85,7% 87,1%

low middle 2 1,811 ,150 14,3% 12,9%

high 0 ,000 ,000 0,0% 0,0%

low 17 17,135 -,076 81,0% 81,6%

low middle middle 3 2,865 ,086 14,3% 13,6%

high 1 1,000 ,000 4,8% 4,8%

high
low 29 28,676 ,167 87,9% 86,9%

upper middle middle 2 2,324 -,220 6,1% 7,0%

high 2 2,000 ,000 6,1% 6,1%

low 32 32,000 ,000 100,0% 100,0%

high middle 0 ,000 ,000 0,0% 0,0%

high 0 ,000 ,000 0,0% 0,0%

The percentages are based on total observed frequencies in each subpopulation.