Alan Taylor

PEMAKAIAN PROSEDUR GLM DALAM IBM SPSS STATISTICS

Oleh :

Abdullah M. Jaubah

Pendahuluan

Penulis, dalam beberapa tahun yang telah lalu, mempelajari bahan-bahan yang terkandung

dalam UCLA Academic Technology Services. Bahan-bahan studi dan penghayatan SPSS ini

banyak sekali mengandung sintaksis sehingga penulis mempelajari sintaksis SPSS tersebut

dan ternyata lebih tangguh daripada cara point and click. Cara pemrograman dan cara point

and click sangat berhubungan sehingga kedua cara itu perlu dipakai sebagai bahan studi dan

penghayatan mengenai IBM SPSS Statistics. Pembahasan atas bahan-bahan tersebut

didominasi oleh cara pemrograman dengan memanfaatkan bahasa perintah sebagaimana

tercermin dalam sintaksis IBM SPSS Statistics. Studi dan penghayatan lebih lanjut diarahkan

pada studi dan penghayatan atas Command Syntax Reference.

Beberapa kritik telah dilancarkan terhadap para penulis buku SPSS. Salah satu kritik adalah

kelangkaan pembahasan sintaksis dalam buku-buku tersebut. Keadaan ini omberbeda dengan

makalah dan buku yang ditulis di Amerika Serikat. Alan Taylor (2002-2011) telah

menyajikan tulisan berjudul Using the GLM Procedure in SPSS. Tulisan ini terdiri dari 149

halaman dan inti dari pembahasan ini tercermin dalam sintaksis yang dipakainya.

Alan Taylor telah menyajikan pokok-pokok pembahasan mengenai : General, One-way

Independent Groups ANOVA, One-Way Indepencent Groups Analysis of Covariance,

Factorial Independent Groups ANOVA, Introduction to Repeated Measures Analyses, One-

way Repeated Measures (Within-subject) Analysis, Two Within-Subjects Factors, A Mixed

Design: One Within- and One Between-Subject Factor, A Numeric Independent Variable in a

Repeated Measures Design, Analysing a Repeated Measures Design with Stacked Data,

Carrying Out Regression Analyses with GLM, A Multiple Regression Analysis with GLM,

Interactions Among Categorical Variables, The Interaction Between a Categorical Variable

and a Numeric Variable, dan pokok pembahasan mengenai The Interaction Between Two

Numeric Variables

1

Contoh sintaksis yang disajikan di sini diambil dari tulisan tersebut yang disajikan secara

terpencar dalam pembahasannya. Sintaksis ini merupakan hasil dari studi dan penghayatan

atas isi tulisan di atas.

Data

Data yang dipakai oleh Alan Taylor dinamakan Alan glmdemo.sav. Arsip data ini akan

disajikan sebagai lampiran.

Sintaksis Dari Alan Taylor

Sintaksis yang disajikan oleh Alan Taylor dikumpulkan di sini. Sintaksis ini adalah sebagai

berikut :

********************************************************** Using the GLM Procedure in SPSS***** Alan Taylor***** Dikumpulkan oleh Abdullah M. Jaubah*****************************************************

GET FILE='D:\ABC\glmdemo.sav'.

*********************************************************** GLM Univariate******************************************************glm pers1 by group with age.

glm pers1 by group with age/method=sstype(3).

glm pers1 by gender group with age.

glm pers1 by gender group with age/design=age gender group age*gender age*group.

glm test2 by group/print=descriptives/plot=profile(group).

glm test2 by group/posthoc=group (lsd bonferroni).

glm test2 by group/contrast(group)=simple(1).

glm test2 by group/contrast(group)=special( -3 1 1 1, -1 -1 1 1, -1 -1 2 0)/print=test(lmatrix).

glm test2 by group/emmeans=table(group) compare(group) adjust(bonferroni).

glm test2 by group/print=parameters.

glm test2 with g2 g3 g4/print=parameters.

2

glm test2 by group with pers2/print=descriptives etasq/emmeans=table(group).

glm test2 by group with pers2/method=sstype(1)/print=etasq.

glm test2 by group with pers2/method=sstype(1)/design = group pers2.

crosstabs gender by group/statistics=chisq.

glm test2 by gender group/plot=profile(group*gender).

glm test2 by gender group/plot=profile(group*gender)/design=gender group.

glm test2 by gender group/print=descriptives parameters/emmeans=table(group) compare(group)/emmeans=table(gender*group)/posthoc=group(lsd)/design=gender group.

glm test2 by gender group/emmeans=table(gender*group) compare(group)/emmeans=table(gender*group) compare(gender).

glm test2 by gender group/print=descriptives/lmatrix="g*gp contrasts" gender*group 1 -1 0 0 -1 1 0 0;gender*group 0 1 0 -1 0 -1 0 1; gender*group 1 -1/3 -1/3 -1/3 -1 +1/3 +1/3 + 1/3.

*********************************************************** GLM Repeated Measure******************************************************

t-test pairs=test1 test2.

glm test1 test2/wsfactor test 2/plot=profile(test).

glm test1 test2 test3/wsfactor test 3/print=rsscp test(mmatrix)/plot=profile(test).

glm test1 test2 test3/wsfactor test 3 helmert/emmeans=table(test) compare(test)/mmatrix="t1vt23" test1 2 test2 -1 test3 -1; "t2vt3" test1 0 test2 1 test3 -1/print=test(mmatrix).

glm site1h1 site1h2 site2h1 site2h2 site3h1 site3h2/wsfactor site 3 simple (1) hemi 2 simple (1)/plot=profile(site*hemi)/print=etasq test(mmatrix).

glm site1 site2 site3 by grp/wsfactor site 3/plot=profile(site*grp).

3

glm site1 site2 site3 by grp/wsfactor site 3 difference/contrast(grp)=simple(1).

glm site1 site2 site3 by grp/wsfactor site 3/print=descriptives/lmatrix "grp2 v 1" grp -1 1/mmatrix "site2 v 1" site1 -1 site2 1 site3 0; "site3 v12" site1 -.5 site2 -.5 site3 1.

glm test1 test3 with pers1/wsfactor test 2/print=etasq.

compute t31diff=test3 - test1.correlations pers1 t31diff.glm t31diff with pers1/print=parameters

glm test1 test3 by grp/wsfactor test 2.

compute t21diff=test2 - test1.glm t21diff/print=parameters.

glm test1 test2 with pers1/wsfactor test 2.

glm t21diff with pers1/print=parameters.

temporary.compute pers1 = pers1 - 4.glm test1 test2 with pers1/wsfactor test 2.

glm test1 test2 with pers1/wsfactor test 2/emmeans=table(test) compare(test) with(pers1=4).

t-test pairs=test3 pers3 with test1 pers1 (paired).

compute t31diff = test3 - test1.compute p31diff = pers3 - pers1.correlations t31diff p31diff.

sort cases by grp.compute swgrp=swgrp+1.if (lag(grp) eq 1 and grp eq 2)swgrp=1.leave swgrp.execute.

select if (swgrp le 20).

varstocases / make resp from site1 site2 site3/index = site(3)//keep = id swgrp grp.

mixed resp by site grp/fixed=intercept site grp site*grp/repeated=site | subject(id) covtype(un)/print=solution testcov.

Hasil Pelaksanaan Sintaksis

4

Hasil pelaksanaan sintaksis di atas adalah sebagai berikut :

********************************************************** Using the GLM Procedure in SPSS***** Alan Taylor***** Dikumpulkan oleh Abdullah M. Jaubah*****************************************************

GET FILE='D:\ABC\glmdemo.sav'.

*********************************************************** GLM Univariate******************************************************glm pers1 by group with age.

glm pers1 by group with age/method=sstype(3).

Between-Subjects Factors

Value Label N

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15

Tests of Between-Subjects Effects

Dependent Variable: pers1

Source Type III Sum of Squares df Mean Square F Sig.

Corrected Model 5.492a 4 1.373 2.504 .047

Intercept 205.249 1 205.249 374.233 .000

age .002 1 .002 .003 .953

group 5.412 3 1.804 3.290 .024

Error 51.554 94 .548

Total 2433.616 99

Corrected Total 57.047 98

a. R Squared = .096 (Adjusted R Squared = .058)

glm pers1 by gender group with age.


Value Label N

gender1 male 57

2 female 42

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15

5




Corrected Model 7.856a 8 .982 1.797 .088

Intercept 195.384 1 195.384 357.479 .000

age .111 1 .111 .204 .653

gender .251 1 .251 .460 .500

group 4.816 3 1.605 2.937 .038

gender * group 2.237 3 .746 1.364 .259

Error 49.191 90 .547

Total 2433.616 99


a. R Squared = .138 (Adjusted R Squared = .061)glm pers1 by gender group with age/design=age gender group age*gender age*group.


Value Label N

gender1 male 57

2 female 42

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15




Corrected Model 7.061a 9 .785 1.397 .202

Intercept 179.543 1 179.543 319.679 .000

age .007 1 .007 .012 .914

gender .363 1 .363 .647 .423

group .290 3 .097 .172 .915

gender * age .295 1 .295 .526 .470

group * age 1.173 3 .391 .696 .557

Error 49.986 89 .562

Total 2433.616 99



glm test2 by group/print=descriptives/plot=profile(group).

6


Value Label N

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15

Descriptive Statistics

Dependent Variable: test2

group Mean Std. Deviation N

1 Control 4.52 .635 29

2 25 mg/kg 4.89 .580 24

3 50 mg/kg 5.12 .814 31

4 100 mg/kg 4.98 1.159 15

Total 4.87 .804 99





Intercept 2175.073 1 2175.073 3581.473 .000

group 5.624 3 1.875 3.087 .031

Error 57.695 95 .607

Total 2410.015 99



7

glm test2 by group/posthoc=group (lsd bonferroni).


Value Label N

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15





Intercept 2175.073 1 2175.073 3581.473 .000

group 5.624 3 1.875 3.087 .031

Error 57.695 95 .607

Total 2410.015 99



8

Multiple Comparisons


(I) (J) Mean Differenc

e (I-J)

Std. Error Sig. 95% Confidence Interval

Lower Bound

Upper Bound

LSD

1 Control

1 Control

2 25 mg/kg -0.36 0.215 0.095 -0.79 0.06

3 50 mg/kg -.60* 0.201 0.004 -1 -0.2

4 100 mg/kg -0.46 0.248 0.069 -0.95 0.04

2 25 mg/kg

1 Control 0.36 0.215 0.095 -0.06 0.79

2 25 mg/kg

3 50 mg/kg -0.24 0.212 0.269 -0.66 0.19

4 100 mg/kg -0.09 0.256 0.718 -0.6 0.42

3 50 mg/kg

1 Control .60* 0.201 0.004 0.2 1

2 25 mg/kg 0.24 0.212 0.269 -0.19 0.66

3 50 mg/kg 4 100 mg/kg 0.14 0.245 0.562 -0.34 0.63

4 100 mg/kg

1 Control 0.46 0.248 0.069 -0.04 0.95

2 25 mg/kg 0.09 0.256 0.718 -0.42 0.6

3 50 mg/kg -0.14 0.245 0.562 -0.63 0.34

4 100 mg/kg

Bonferroni

1 Control

1 Control

2 25 mg/kg -0.36 0.215 0.572 -0.94 0.22

3 50 mg/kg -.60* 0.201 0.023 -1.14 -0.06

4 100 mg/kg -0.46 0.248 0.416 -1.12 0.21

2 25 mg/kg

1 Control 0.36 0.215 0.572 -0.22 0.94

2 25 mg/kg 3 50 mg/kg -0.24 0.212 1 -0.81 0.34

4 100 mg/kg -0.09 0.256 1 -0.78 0.6

3 50 mg/kg

1 Control .60* 0.201 0.023 0.06 1.14

2 25 mg/kg 0.24 0.212 1 -0.34 0.81

3 50 mg/kg 4 100 mg/kg 0.14 0.245 1 -0.52 0.8

4 100 mg/kg

1 Control 0.46 0.248 0.416 -0.21 1.12

2 25 mg/kg 0.09 0.256 1 -0.6 0.78

3 50 mg/kg -0.14 0.245 1 -0.8 0.52

4 100 mg/kg Based on observed means.

The error term is Mean Square(Error) = .607.*. The mean difference is significant at the .050 level.

glm test2 by group/contrast(group)=simple(1).


9

Value Label N

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15





Intercept 2175.073 1 2175.073 3581.473 .000

group 5.624 3 1.875 3.087 .031

Error 57.695 95 .607

Total 2410.015 99



Custom Hypothesis Tests

Contrast Results (K Matrix)

10

Simple Contrasta Dependent Variable

test2

Level 2 vs. Level 1

Contrast Estimate .362

Hypothesized Value 0

Difference (Estimate - Hypothesized) .362

Std. Error .215

Sig. .095

95% Confidence Interval for

Difference

Lower Bound -.065

Upper Bound .789

Level 3 vs. Level 1




Std. Error .201

Sig. .004


Difference

Lower Bound .198

Upper Bound .998

Level 4 vs. Level 1




Std. Error .248

Sig. .069


Difference

Lower Bound -.037

Upper Bound .947

a. Reference category = 1

Test Results


Source Sum of Squares df Mean Square F Sig.

Contrast 5.624 3 1.875 3.087 .031

Error 57.695 95 .607

glm test2 by group/contrast(group)=special( -3 1 1 1, -1 -1 1 1, -1 -1 2 0)/print=test(lmatrix).


Value Label N

group 1 Control 29

2 25 mg/kg 24

11

3 50 mg/kg 31

4 100 mg/kg 15





Intercept 2175.073 1 2175.073 3581.473 .000

group 5.624 3 1.875 3.087 .031

Error 57.695 95 .607

Total 2410.015 99



Contrast Coefficients (L' Matrix)

Intercept

Parameter Contrast

L1

Intercept 1

[group=1] .250

[group=2] .250

[group=3] .250

[group=4] .250

The default display of this matrix is the transpose of the corresponding L matrix.

Based on Type III Sums of Squares.

group

Parameter Contrast

L2 L3 L4

Intercept 0 0 0

[group=1] 1 0 0

[group=2] 0 1 0

[group=3] 0 0 1

[group=4] -1 -1 -1

The default display of this matrix is the transpose of the corresponding L matrix.

Based on Type III Sums of Squares.


12

Contrast Coefficients (L' Matrix)

Parameter Special Contrast

L1 L2 L3

Intercept 0 0 0

[group=1] -3 -1 -1

[group=2] 1 -1 -1

[group=3] 1 1 2

[group=4] 1 1 0

The default display of this matrix is the transpose of the

corresponding L matrix.


Special Contrast Dependent Variable

test2

L1

Contrast Estimate 1.416


Difference (Estimate - Hypothesized) 1.416

Std. Error .523

Sig. .008


Difference

Lower Bound .377

Upper Bound 2.454

L2




Std. Error .326

Sig. .037


Difference

Lower Bound .044

Upper Bound 1.338

L3




Std. Error .353

Sig. .020


Difference

Lower Bound .133

Upper Bound 1.534

13

Test Results



Contrast 5.624 3 1.875 3.087 .031

Error 57.695 95 .607

glm test2 by group/emmeans=table(group) compare(group) adjust(bonferroni).


Value Label N

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15





Intercept 2175.073 1 2175.073 3581.473 .000

group 5.624 3 1.875 3.087 .031

Error 57.695 95 .607

Total 2410.015 99



Estimated Marginal Means

Estimates


Mean Std. Error 95% Confidence Interval

Lower Bound Upper Bound

1 Control 4.525 .145 4.237 4.812

2 25 mg/kg 4.887 .159 4.571 5.203

3 50 mg/kg 5.123 .140 4.845 5.400

14

4 100 mg/kg 4.980 .201 4.581 5.379

Pairwise Comparisons


(I) (J) Mean

Difference

(I-J)

Std. Error Sig.b 95% Confidence Interval for Differenceb


1 Control

1 Control

2 25 mg/kg -.362 .215 .572 -.942 .217

3 50 mg/kg -.598* .201 .023 -1.140 -.055

4 100 mg/kg -.455 .248 .416 -1.123 .213

2 25 mg/kg

1 Control .362 .215 .572 -.217 .942

2 25 mg/kg

3 50 mg/kg -.236 .212 1.000 -.807 .335

4 100 mg/kg -.093 .256 1.000 -.784 .598

3 50 mg/kg

1 Control .598* .201 .023 .055 1.140

2 25 mg/kg .236 .212 1.000 -.335 .807

3 50 mg/kg

4 100 mg/kg .143 .245 1.000 -.518 .803

4 100 mg/kg

1 Control .455 .248 .416 -.213 1.123

2 25 mg/kg .093 .256 1.000 -.598 .784

3 50 mg/kg -.143 .245 1.000 -.803 .518

4 100 mg/kg

Based on estimated marginal means

*. The mean difference is significant at the .050 level.

b. Adjustment for multiple comparisons: Bonferroni.

Univariate Tests


Sum of Squares df Mean Square F Sig.

Contrast 5.624 3 1.875 3.087 .031

Error 57.695 95 .607

The F tests the effect of . This test is based on the linearly independent pairwise comparisons among

the estimated marginal means.

glm test2 by group/print=parameters.


Value Label N

15

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15





Intercept 2175.073 1 2175.073 3581.473 .000

group 5.624 3 1.875 3.087 .031

Error 57.695 95 .607

Total 2410.015 99



Parameter Estimates


Parameter B Std. Error t Sig. 95% Confidence Interval


Intercept 4.980 .201 24.750 .000 4.581 5.379

[group=1] -.455 .248 -1.837 .069 -.947 .037

[group=2] -.093 .256 -.363 .718 -.602 .416

[group=3] .143 .245 .582 .562 -.344 .629

[group=4] 0a . . . . .

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

glm test2 by group with pers2/print=descriptives etasq/emmeans=table(group).


Value Label N

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15



group Mean Std. Deviation N

16

1 Control 4.52 .635 29

2 25 mg/kg 4.89 .580 24

3 50 mg/kg 5.12 .814 31

4 100 mg/kg 4.98 1.159 15

Total 4.87 .804 99



Source Type III Sum of

Squares

df Mean

Square

F Sig. Partial Eta Squared

Corrected Model 7.588a 4 1.897 3.200 .016 .120

Intercept 35.562 1 35.562 59.981 .000 .390

pers2 1.964 1 1.964 3.313 .072 .034

group 3.467 3 1.156 1.949 .127 .059

Error 55.731 94 .593

Total 2410.015 99







1 Control 4.593 .148 4.300 4.887

2 25 mg/kg 4.873a .157 4.560 5.185

3 50 mg/kg 5.089a .140 4.812 5.366

4 100 mg/kg 4.941a .200 4.543 5.338

a. Covariates appearing in the model are evaluated at the following values: pers2 = 4.96.

glm test2 by group with pers2/method=sstype(1)/print=etasq.


Value Label N

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15

17



Source Type I Sum of

Squares

df Mean Square F Sig. Partial Eta Squared

Corrected Model 7.588a 4 1.897 3.200 .016 .120

Intercept 2346.696 1 2346.696 3958.139 .000 .977

pers2 4.121 1 4.121 6.952 .010 .069

group 3.467 3 1.156 1.949 .127 .059

Error 55.731 94 .593

Total 2410.015 99



glm test2 by group with pers2/method=sstype(1)/design = group pers2.


Value Label N

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15



Source Type I Sum of Squares df Mean Square F Sig.


Intercept 2346.696 1 2346.696 3958.139 .000

group 5.624 3 1.875 3.162 .028

pers2 1.964 1 1.964 3.313 .072

Error 55.731 94 .593

Total 2410.015 99



18

crosstabs gender by group/statistics=chisq.

Case Processing Summary

Cases

Valid Missing Total

N Percent N Percent N Percent

gender * group 99 100.0% 0 0.0% 99 100.0%

gender * group Crosstabulation

Count

group Total

1 Control 2 25 mg/kg 3 50 mg/kg 4 100 mg/kg

gender1 male 18 16 14 9 57

2 female 11 8 17 6 42

Total 29 24 31 15 99

Chi-Square Tests

Value df Asymp. Sig. (2-sided)

Pearson Chi-Square 3.044a 3 .385

Continuity Correction

Likelihood Ratio 3.038 3 .386

Linear-by-Linear Association .723 1 .395

N of Valid Cases 99

a. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 6.36.

glm test2 by gender group/plot=profile(group*gender).


Value Label N

gender1 male 57

2 female 42

group 1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

19

4 100 mg/kg 15





Intercept 2065.947 1 2065.947 3437.003 .000

gender .948 1 .948 1.577 .212

group 5.038 3 1.679 2.794 .045

gender * group 1.507 3 .502 .835 .478

Error 54.699 91 .601

Total 2410.015 99



Profile Plots

20

glm test2 by gender group/plot=profile(group*gender)/design=gender group.


Value Label N

gender1 male 57

2 female 42

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15





Intercept 2134.395 1 2134.395 3569.619 .000

21

gender 1.489 1 1.489 2.490 .118

group 4.938 3 1.646 2.753 .047

Error 56.206 94 .598

Total 2410.015 99



glm test2 by gender group/print=descriptives parameters/emmeans=table(group) compare(group)/emmeans=table(gender*group)/posthoc=group(lsd)/design=gender group.


Value Label N

gender1 male 57

2 female 42

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15



gender group Mean Std. Deviation N

1 male 1 Control 4.49 .540 18

2 25 mg/kg 4.71 .535 16

3 50 mg/kg 4.90 .589 14

22

4 100 mg/kg 5.06 1.114 9

Total 4.74 .683 57

2 female

1 Control 4.59 .792 11

2 25 mg/kg 5.25 .520 8

3 50 mg/kg 5.30 .939 17

4 100 mg/kg 4.86 1.320 6

Total 5.04 .924 42

Total

1 Control 4.52 .635 29

2 25 mg/kg 4.89 .580 24

3 50 mg/kg 5.12 .814 31

4 100 mg/kg 4.98 1.159 15

Total 4.87 .804 99





Intercept 2134.395 1 2134.395 3569.619 .000

gender 1.489 1 1.489 2.490 .118

group 4.938 3 1.646 2.753 .047

Error 56.206 94 .598

Total 2410.015 99



Parameter Estimates




Intercept 5.131 .221 23.169 .000 4.691 5.571

[gender=1] -.252 .160 -1.578 .118 -.569 .065

[gender=2] 0a . . . . .

[group=1] -.450 .246 -1.830 .070 -.938 .038

[group=2] -.076 .255 -.299 .765 -.582 .430

[group=3] .105 .244 .430 .668 -.380 .590

[group=4] 0a . . . . .

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


Estimates

23




1 Control 4.555 .145 4.267 4.843

2 25 mg/kg 4.929 .160 4.611 5.247

3 50 mg/kg 5.110 .139 4.834 5.387

4 100 mg/kg 5.005 .200 4.607 5.403



(I) (J) Mean Difference (I-J) Std.

Error

Sig.b 95% Confidence Interval for Differenceb


1 Control

1 Control

2 25 mg/kg -.374 .214 .083 -.798 .050

3 50 mg/kg -.555* .202 .007 -.956 -.155

4 100 mg/kg -.450 .246 .070 -.938 .038

2 25 mg/kg

1 Control .374 .214 .083 -.050 .798

2 25 mg/kg

3 50 mg/kg -.181 .213 .397 -.604 .242

4 100 mg/kg -.076 .255 .765 -.582 .430

3 50 mg/kg

1 Control .555* .202 .007 .155 .956

2 25 mg/kg .181 .213 .397 -.242 .604

3 50 mg/kg

4 100 mg/kg .105 .244 .668 -.380 .590

4 100 mg/kg 1 Control .450 .246 .070 -.038 .938

2 25 mg/kg .076 .255 .765 -.430 .582

24

3 50 mg/kg -.105 .244 .668 -.590 .380

4 100 mg/kg



b. Adjustment for multiple comparisons: Least Significant Difference (equivalent to no adjustments).

Univariate Tests



Contrast 4.938 3 1.646 2.753 .047

Error 56.206 94 .598

The F tests the effect of . This test is based on the linearly independent pairwise comparisons among


2. *




1 male

1 Control 4.429 .156 4.120 4.738

2 25 mg/kg 4.803 .167 4.472 5.134

3 50 mg/kg 4.984 .164 4.658 5.310

4 100 mg/kg 4.879 .210 4.463 5.295

2 female

1 Control 4.681 .174 4.335 5.028

2 25 mg/kg 5.055 .190 4.677 5.433

3 50 mg/kg 5.236 .156 4.926 5.547

4 100 mg/kg 5.131 .221 4.691 5.571

Post Hoc Tests

Multiple Comparisons


LSD

(I) (J) Mean Difference (I- Std. Sig. 95% Confidence Interval

25

J) Error Lower Bound Upper Bound

1 Control

1 Control

2 25 mg/kg -.36 .213 .093 -.79 .06

3 50 mg/kg -.60* .200 .004 -.99 -.20

4 100 mg/kg -.46 .246 .067 -.94 .03

2 25 mg/kg

1 Control .36 .213 .093 -.06 .79

2 25 mg/kg

3 50 mg/kg -.24 .210 .265 -.65 .18

4 100 mg/kg -.09 .255 .716 -.60 .41

3 50 mg/kg

1 Control .60* .200 .004 .20 .99

2 25 mg/kg .24 .210 .265 -.18 .65

3 50 mg/kg

4 100 mg/kg .14 .243 .559 -.34 .63

4 100 mg/kg

1 Control .46 .246 .067 -.03 .94

2 25 mg/kg .09 .255 .716 -.41 .60

3 50 mg/kg -.14 .243 .559 -.63 .34

4 100 mg/kg

Based on observed means.

The error term is Mean Square(Error) = .598.


glm test2 by gender group/emmeans=table(gender*group) compare(group)/emmeans=table(gender*group) compare(gender).


Value Label N

gender1 male 57

2 female 42

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15





Intercept 2065.947 1 2065.947 3437.003 .000

gender .948 1 .948 1.577 .212

group 5.038 3 1.679 2.794 .045

gender * group 1.507 3 .502 .835 .478

26

Error 54.699 91 .601

Total 2410.015 99




Estimates




1 male

1 Control 4.487 .183 4.124 4.850

2 25 mg/kg 4.707 .194 4.322 5.092

3 50 mg/kg 4.902 .207 4.491 5.314

4 100 mg/kg 5.060 .258 4.547 5.573

2 female

1 Control 4.586 .234 4.121 5.050

2 25 mg/kg 5.246 .274 4.702 5.790

3 50 mg/kg 5.304 .188 4.930 5.677

4 100 mg/kg 4.860 .317 4.231 5.489


Dependent Variable: test2 (I) (J) Mean

Difference (I-J)

Std. Error Sig.b 95% Confidence Interval for Differenceb

Lower Bound

Upper Bound

1 male

1 Control

1 Control

2 25 mg/kg -0.22 0.266 0.411 -0.749 0.309

3 50 mg/kg -0.415 0.276 0.137 -0.964 0.134

4 100 mg/kg -0.572 0.317 0.074 -1.201 0.056

2 25 mg/kg

1 Control 0.22 0.266 0.411 -0.309 0.749

2 25 mg/kg 3 50 mg/kg -0.195 0.284 0.494 -0.758 0.3694 100 mg/kg -0.352 0.323 0.278 -0.994 0.289

3 50 mg/kg

1 Control 0.415 0.276 0.137 -0.134 0.964

2 25 mg/kg 0.195 0.284 0.494 -0.369 0.758

3 50 mg/kg 4 100 mg/kg -0.158 0.331 0.635 -0.816 0.5

4 100 mg/kg

1 Control 0.572 0.317 0.074 -0.056 1.2012 25 mg/kg 0.352 0.323 0.278 -0.289 0.994

3 50 mg/kg 0.158 0.331 0.635 -0.5 0.816

4 100 mg/kg

27

2 female

1 Control

1 Control 2 25 mg/kg -0.66 0.36 0.07 -1.376 0.055

3 50 mg/kg -.718* 0.3 0.019 -1.314 -0.123

4 100 mg/kg -0.275 0.393 0.487 -1.056 0.507

2 25 mg/kg

1 Control 0.66 0.36 0.07 -0.055 1.376

2 25 mg/kg 3 50 mg/kg -0.058 0.332 0.862 -0.718 0.6024 100 mg/kg 0.386 0.419 0.359 -0.446 1.218

3 50 mg/kg

1 Control .718* 0.3 0.019 0.123 1.314

2 25 mg/kg 0.058 0.332 0.862 -0.602 0.718

3 50 mg/kg

4 100 mg/kg 0.444 0.368 0.231 -0.287 1.175

4 100 mg/kg

1 Control 0.275 0.393 0.487 -0.507 1.056

2 25 mg/kg -0.386 0.419 0.359 -1.218 0.446

3 50 mg/kg -0.444 0.368 0.231 -1.175 0.287

4 100 mg/kg

Based on estimated marginal means*. The mean difference is significant at the .050 level.




1 maleContrast 2.455 3 .818 1.361 .260

Error 54.699 91 .601

2 femaleContrast 3.990 3 1.330 2.213 .092

Error 54.699 91 .601

Each F tests the simple effects of within each level combination of the other effects shown. These tests are based on the

linearly independent pairwise comparisons among the estimated marginal means.

Estimates




1 male

1 Control 4.487 .183 4.124 4.850

2 25 mg/kg 4.707 .194 4.322 5.092

3 50 mg/kg 4.902 .207 4.491 5.314

4 100 mg/kg 5.060 .258 4.547 5.573

28

2 female

1 Control 4.586 .234 4.121 5.050

2 25 mg/kg 5.246 .274 4.702 5.790

3 50 mg/kg 5.304 .188 4.930 5.677

4 100 mg/kg 4.860 .317 4.231 5.489

Univariate Tests



1 ControlContrast .066 1 .066 .109 .742

Error 54.699 91 .601

2 25 mg/kgContrast 1.547 1 1.547 2.574 .112

Error 54.699 91 .601

3 50 mg/kgContrast 1.239 1 1.239 2.061 .155

Error 54.699 91 .601

4 100 mg/kgContrast .144 1 .144 .239 .626

Error 54.699 91 .601

Each F tests the simple effects of within each level combination of the other effects shown. These tests are based on the

linearly independent pairwise comparisons among the estimated marginal means.


Dependent Variable: test2 (I) (J) Mean

Difference (I-J)

Std. Error Sig.a 95% Confidence Interval for Differencea


1 Control

1 male

1 male

2 female -0.098 0.297 0.742 -0.687 0.491

2 female

1 male 0.098 0.297 0.742 -0.491 0.687

2 female

2 25 mg/kg

1 male

1 male

2 female -0.539 0.336 0.112 -1.205 0.128

2 female

1 male 0.539 0.336 0.112 -0.128 1.205

2 female

3 50 1 male 1 male

29

mg/kg

2 female -0.402 0.28 0.155 -0.958 0.154

2 female

1 male 0.402 0.28 0.155 -0.154 0.958

2 female

4 100 mg/kg

1 male

1 male

2 female 0.2 0.409 0.626 -0.612 1.011

2 female

1 male -0.2 0.409 0.626 -1.011 0.612

2 female

Based on estimated marginal meansa. Adjustment for multiple comparisons: Least Significant Difference (equivalent to no adjustments).

glm test2 by gender group/print=descriptives/lmatrix="g*gp contrasts" gender*group 1 -1 0 0 -1 1 0 0;gender*group 0 1 0 -1 0 -1 0 1; gender*group 1 -1/3 -1/3 -1/3 -1 +1/3 +1/3 + 1/3.


Value Label N

gender1 male 57

2 female 42

group

1 Control 29

2 25 mg/kg 24

3 50 mg/kg 31

4 100 mg/kg 15



gender group Mean Std. Deviation N

1 male

1 Control 4.49 .540 18

2 25 mg/kg 4.71 .535 16

3 50 mg/kg 4.90 .589 14

4 100 mg/kg 5.06 1.114 9

Total 4.74 .683 57

2 female

1 Control 4.59 .792 11

2 25 mg/kg 5.25 .520 8

3 50 mg/kg 5.30 .939 17

4 100 mg/kg 4.86 1.320 6

Total 5.04 .924 42

30

Total

1 Control 4.52 .635 29

2 25 mg/kg 4.89 .580 24

3 50 mg/kg 5.12 .814 31

4 100 mg/kg 4.98 1.159 15

Total 4.87 .804 99





Intercept 2065.947 1 2065.947 3437.003 .000

gender .948 1 .948 1.577 .212

group 5.038 3 1.679 2.794 .045

gender * group 1.507 3 .502 .835 .478

Error 54.699 91 .601

Total 2410.015 99




Contrast Results (K Matrix)a

Contrast Dependent Variable

test2

L1




Std. Error .448

Sig. .328


Difference

Lower Bound -.449

Upper Bound 1.330

L2 Contrast Estimate -.738


Difference (Estimate - Hypothesized) -.738

Std. Error .529

31

Sig. .166


Difference

Lower Bound -1.789

Upper Bound .312

L3




Std. Error .358

Sig. .678


Difference

Lower Bound -.561

Upper Bound .859

a. Based on the user-specified contrast coefficients (L') matrix: g*gp contrasts

Test Results



Contrast 1.507 3 .502 .835 .478

Error 54.699 91 .601

*********************************************************** GLM Repeated Measure******************************************************

t-test pairs=test1 test2.

Paired Samples Statistics

Mean N Std. Deviation Std. Error Mean

Pair 1test1 4.61 99 .794 .080

test2 4.87 99 .804 .081

Paired Samples Correlations

N Correlation Sig.

Pair 1 test1 & test2 99 .092 .364

Paired Samples Test

Paired Differences t df Sig. (2-tailed)Mean Std.

DeviationStd. Error

Mean95% Confidence Interval of

the Difference

32

Lower Upper

Pair 1test1 - test2 -0.254 1.076 0.108 -0.468 -0.039 -2.345 98 0.021

glm test1 test2/wsfactor test 2/plot=profile(test).

Within-Subjects Factors

Measure: MEASURE_1

test Dependent Variable

1 test1

2 test2

Multivariate Testsa

Effect Value F Hypothesis df Error df Sig.

test

Pillai's Trace .053 5.500b 1.000 98.000 .021

Wilks' Lambda .947 5.500b 1.000 98.000 .021

Hotelling's Trace .056 5.500b 1.000 98.000 .021

Roy's Largest Root .056 5.500b 1.000 98.000 .021

a. Design: Intercept

Within Subjects Design: test

b. Exact statistic

Mauchly's Test of Sphericitya

Measure: MEASURE_1Within Subjects Effect

Mauchly's W Approx. Chi-

Square

df Sig. Epsilonb

Greenhouse-Geisser

Huynh-Feldt Lower-bound

test 1 0 0 . 1 1 1

Tests the null hypothesis that the error covariance matrix of the orthonormalized transformed dependent variables is proportional to an identity matrix.a. Design: Intercept


b. May be used to adjust the degrees of freedom for the averaged tests of significance. Corrected tests are displayed in the Tests of Within-Subjects Effects table.

33

Tests of Within-Subjects Effects

Measure: MEASURE_1


test

Sphericity Assumed 3.186 1 3.186 5.500 .021

Greenhouse-Geisser 3.186 1.000 3.186 5.500 .021

Huynh-Feldt 3.186 1.000 3.186 5.500 .021

Lower-bound 3.186 1.000 3.186 5.500 .021

Error(test)

Sphericity Assumed 56.772 98 .579

Greenhouse-Geisser 56.772 98.000 .579

Huynh-Feldt 56.772 98.000 .579

Lower-bound 56.772 98.000 .579

Tests of Within-Subjects Contrasts

Measure: MEASURE_1

Source test Type III Sum of Squares df Mean Square F Sig.

test Linear 3.186 1 3.186 5.500 .021

Error(test) Linear 56.772 98 .579


Measure: MEASURE_1

Transformed Variable: Average


Intercept 4452.006 1 4452.006 6388.629 .000

Error 68.293 98 .697

Profile Plots

34

glm test1 test2 test3/wsfactor test 3/print=rsscp test(mmatrix)/plot=profile(test).


Measure: MEASURE_1


1 test1

2 test2

3 test3

Bartlett's Test of Sphericitya

Likelihood Ratio .012

Approx. Chi-Square 8.653

df 5

Sig. .124

Tests the null hypothesis that the residual covariance matrix is proportional to an identity matrix.



Multivariate Testsa


test Pillai's Trace .235 14.938b 2.000 97.000 .000

35

Wilks' Lambda .765 14.938b 2.000 97.000 .000





b. Exact statistic



Mauchly's W Approx. Chi-Square

df Sig. Epsilonb

Greenhouse-Geisser


test 0.993 0.663 2 0.718 0.993 1 0.5





Measure: MEASURE_1


test



Huynh-Feldt 15.681 2.000 7.840 13.896 .000

Lower-bound 15.681 1.000 15.681 13.896 .000

Error(test)



Huynh-Feldt 110.587 196.000 .564

Lower-bound 110.587 98.000 1.128


Measure: MEASURE_1


testLinear 4.703 1 4.703 7.907 .006

Quadratic 10.977 1 10.977 20.572 .000

Error(test)Linear 58.293 98 .595

Quadratic 52.294 98 .534


36

Measure: MEASURE_1



Intercept 6275.753 1 6275.753 7144.781 .000

Error 86.080 98 .878

Transformation Coefficients (M Matrix)

Average

Measure: MEASURE_1

Transformed Variable: AVERAGE

test1 .577

test2 .577

test3 .577

testa

Measure: MEASURE_1

Dependent Variable test

Linear Quadratic

test1 -.707 .408

test2 .000 -.816

test3 .707 .408

a. The contrasts for the within subjects factors are:

test: Polynomial contrast

Residual SSCP Matrix

test1 test2 test3

Sum-of-Squares and Cross-

Products

test1 61.746 5.760 8.381

test2 5.760 63.319 16.645

test3 8.381 16.645 71.603

Covariance

test1 .630 .059 .086

test2 .059 .646 .170

test3 .086 .170 .731

Correlation

test1 1.000 .092 .126

test2 .092 1.000 .247

test3 .126 .247 1.000

Based on Type III Sum of Squares

Profile Plots

37

glm test1 test2 test3/wsfactor test 3 helmert/emmeans=table(test) compare(test)/mmatrix="t1vt23" test1 2 test2 -1 test3 -1; "t2vt3" test1 0 test2 1 test3 -1/print=test(mmatrix).


Measure: MEASURE_1


1 test1

2 test2

3 test3

Multivariate Testsa

38


test

Pillai's Trace .235 14.938b 2.000 97.000 .000

Wilks' Lambda .765 14.938b 2.000 97.000 .000





b. Exact statistic



Mauchly's W

Approx. Chi-

Square

df Sig. Epsilonb

Greenhouse-Geisser

Huynh-Feldt

Lower-bound

test 0.993 0.663 2 0.718 0.993 1 0.5





Measure: MEASURE_1


test



Huynh-Feldt 15.681 2.000 7.840 13.896 .000

Lower-bound 15.681 1.000 15.681 13.896 .000

Error(test)



Huynh-Feldt 110.587 196.000 .564

Lower-bound 110.587 98.000 1.128


Measure: MEASURE_1

39


testLevel 1 vs. Later .074 1 .074 .080 .777

Level 2 vs. Level 3 31.263 1 31.263 30.146 .000

Error(test)Level 1 vs. Later 89.658 98 .915

Level 2 vs. Level 3 101.631 98 1.037


Measure: MEASURE_1



Intercept 2091.918 1 2091.918 7144.781 .000

Error 28.693 98 .293


Average

Measure: MEASURE_1


test1 .333

test2 .333

test3 .333

testa

Measure: MEASURE_1


Level 1 vs. Later Level 2 vs. Level 3

test1 1.000 .000

test2 -.500 1.000

test3 -.500 -1.000


test: Helmert contrast


40


Dependent Variable Transformed Variable

t1vt23 t2vt3

test1 2.000 .000

test2 -1.000 1.000

test3 -1.000 -1.000


Contrasta Transformed Variable

t1vt23 t2vt3

L1

Contrast Estimate .055 .562

Hypothesized Value 0 0

Difference (Estimate - Hypothesized) .055 .562

Std. Error .192 .102

Sig. .777 .000


Difference

Lower Bound -.327 .359

Upper Bound .436 .765

a. Estimable Function for Intercept

Multivariate Test Results

Value F Hypothesis df Error df Sig.

Pillai's trace .235 14.938a 2.000 97.000 .000

Wilks' lambda .765 14.938a 2.000 97.000 .000

Hotelling's trace .308 14.938a 2.000 97.000 .000

Roy's largest root .308 14.938a 2.000 97.000 .000

a. Exact statistic

Univariate Test Results

Source Transformed Variable Sum of Squares df Mean Square F Sig.

Contrastt1vt23 .294 1 .294 .080 .777

t2vt3 31.263 1 31.263 30.146 .000

Errort1vt23 358.631 98 3.660

t2vt3 101.631 98 1.037


41

test


Measure: MEASURE_1


1 2 3

test1 1 0 0

test2 0 1 0

test3 0 0 1

Estimates

Measure: MEASURE_1

test Mean Std. Error 95% Confidence Interval


1 4.615 .080 4.457 4.773

2 4.869 .081 4.708 5.029

3 4.307 .086 4.136 4.477


Measure: MEASURE_1

(I) test (J) test Mean Difference (I-J) Std. Error Sig.b 95% Confidence Interval for Differenceb


1

1

2 -.254* .108 .021 -.468 -.039

3 .308* .110 .006 .091 .526

2

1 .254* .108 .021 .039 .468

2

3 .562* .102 .000 .359 .765

3

1 -.308* .110 .006 -.526 -.091

2 -.562* .102 .000 -.765 -.359

3




Multivariate Tests


42

Pillai's trace .235 14.938a 2.000 97.000 .000

Wilks' lambda .765 14.938a 2.000 97.000 .000



Each F tests the multivariate effect of test. These tests are based on the linearly independent pairwise

comparisons among the estimated marginal means.

a. Exact statistic

glm site1h1 site1h2 site2h1 site2h2 site3h1 site3h2/wsfactor site 3 simple (1) hemi 2 simple (1)/plot=profile(site*hemi)/print=etasq test(mmatrix).


Measure: MEASURE_1

site hemi Dependent Variable

11 site1h1

2 site1h2

21 site2h1

2 site2h2

31 site3h1

2 site3h2

Multivariate Testsa

Effect Value F Hypothesis df Error df Sig. Partial Eta Squared

site

Pillai's Trace 0.537 56.245b 2 97 0 0.537

Wilks' Lambda 0.463 56.245b 2 97 0 0.537

Hotelling's Trace 1.16 56.245b 2 97 0 0.537

Roy's Largest Root 1.16 56.245b 2 97 0 0.537

hemi

Pillai's Trace 0.051 5.269b 1 98 0.024 0.051

Wilks' Lambda 0.949 5.269b 1 98 0.024 0.051

Hotelling's Trace 0.054 5.269b 1 98 0.024 0.051

Roy's Largest Root 0.054 5.269b 1 98 0.024 0.051

site * hemi

Pillai's Trace 0.209 12.797b 2 97 0 0.209

Wilks' Lambda 0.791 12.797b 2 97 0 0.209

Hotelling's Trace 0.264 12.797b 2 97 0 0.209

Roy's Largest Root 0.264 12.797b 2 97 0 0.209a. Design: Intercept

Within Subjects Design: site + hemi + site * hemi

b. Exact statistic


Measure: MEASURE_1Within Mauchly's W Approx. df Sig. Epsilonb

43

Subjects Effect

Chi-Square Greenhouse-Geisser Huynh-Feldt Lower-bound

site 0.704 34.046 2 0 0.772 0.782 0.5

hemi 1 0 0 . 1 1 1

site * hemi 0.818 19.436 2 0 0.846 0.86 0.5


Within Subjects Design: site + hemi + site * hemi



Measure: MEASURE_1

Source Type III Sum of Squares

df Mean Square


site

Sphericity Assumed 207.363 2 103.681 82.995 0 0.459

Greenhouse-Geisser 207.363 1.543 134.372 82.995 0 0.459

Huynh-Feldt 207.363 1.563 132.657 82.995 0 0.459

Lower-bound 207.363 1 207.363 82.995 0 0.459

Error(site)

Sphericity Assumed 244.852 196 1.249 Greenhouse-Geisser 244.852 151.23 1.619 Huynh-Feldt 244.852 153.19 1.598 Lower-bound 244.852 98 2.498

hemi

Sphericity Assumed 5.446 1 5.446 5.269 0.024 0.051

Greenhouse-Geisser 5.446 1 5.446 5.269 0.024 0.051

Huynh-Feldt 5.446 1 5.446 5.269 0.024 0.051

Lower-bound 5.446 1 5.446 5.269 0.024 0.051

Error(hemi)

Sphericity Assumed 101.286 98 1.034 Greenhouse-Geisser 101.286 98 1.034 Huynh-Feldt 101.286 98 1.034 Lower-bound 101.286 98 1.034

site * hemi

Sphericity Assumed 27.821 2 13.91 17.834 0 0.154

Greenhouse-Geisser 27.821 1.693 16.436 17.834 0 0.154

Huynh-Feldt 27.821 1.719 16.182 17.834 0 0.154

Lower-bound 27.821 1 27.821 17.834 0 0.154

Error(site*hemi)

Sphericity Assumed 152.876 196 0.78 Greenhouse-Geisser 152.876 165.88 0.922 Huynh-Feldt 152.876 168.48 0.907 Lower-bound 152.876 98 1.56


Measure: MEASURE_1

44

Source site hemi Type III Sum of Squares

df Mean Square


site

Level 2 vs. Level 1

6.303 1 6.303 8.842 0.004 0.083

Level 2 vs. Level 1

Level 3 vs. Level 1

183.2 1 183.201 97.502 0 0.499

Level 2 vs. Level 1

Level 2 vs. Level 1

Error(site)

Level 2 vs. Level 1

69.86 98 0.713

Level 2 vs. Level 1

Level 3 vs. Level 1

184.14 98 1.879

Level 2 vs. Level 1

Level 2 vs. Level 1

hemi

Level 2 vs. Level 1

Level 2 vs. Level 1

Level 3 vs. Level 1

Level 2 vs. Level 1

Level 2 vs. Level 1 3.631 1 3.631 5.269 0.024 0.051

Error(hemi)

Level 2 vs. Level 1

Level 2 vs. Level 1

Level 3 vs. Level 1

Level 2 vs. Level 1

Level 2 vs. Level 1 67.524 98 0.689

site * hemi

Level 2 vs. Level 1

Level 2 vs. Level 1 100.79 1 100.792 25.729 0 0.208

Level 3 vs. Level 1

Level 2 vs. Level 1 4.905 1 4.905 2.725 0.102 0.027

Level 2 vs. Level 1

Error(site*hemi)

Level 2 vs. Level 1

Level 2 vs. Level 1 383.91 98 3.917

Level 3 vs. Level 1

Level 2 vs. Level 1 176.4 98 1.8

Level 2 vs. Level 1


Measure: MEASURE_1


45

Source Type III Sum of Squares df Mean

Square

F Si

g.

Partial Eta

Squared

Intercept 1352.250 1 1352.2505167.7

50

.0

00.981

Error 25.644 98 .262


Average

Measure: MEASURE_1


site1h1 .167

site1h2 .167

site2h1 .167

site2h2 .167

site3h1 .167

site3h2 .167

sitea

Measure: MEASURE_1

Dependent Variable site

Level 2 vs. Level 1 Level 3 vs. Level 1

site1h1 -.500 -.500

site1h2 -.500 -.500

site2h1 .500 .000

site2h2 .500 .000

site3h1 .000 .500

site3h2 .000 .500


site: Simple contrast

hemia

Measure: MEASURE_1

46

Dependent Variable hemi

Level 2 vs. Level 1

site1h1 -.333

site1h2 .333

site2h1 -.333

site2h2 .333

site3h1 -.333

site3h2 .333


hemi: Simple contrast

site * hemia

Measure: MEASURE_1

Dependent Variable site


hemi hemi


site1h1 1 1

site1h2 -1 -1

site2h1 -1 0

site2h2 1 0

site3h1 0 -1

site3h2 0 1


site: Simple contrast

hemi: Simple contrast

47

glm site1 site2 site3 by grp/wsfactor site 3/plot=profile(site*grp).


Measure: MEASURE_1

site Dependent Variable

1 site1

2 site2

3 site3


Value Label N

grp1 Control 51

2 Experimental 48

48

Multivariate Testsa


site

Pillai's Trace .542 56.803b 2.000 96.000 .000

Wilks' Lambda .458 56.803b 2.000 96.000 .000

Hotelling's Trace 1.183 56.803b 2.000 96.000 .000

Roy's Largest Root 1.183 56.803b 2.000 96.000 .000

site * grp

Pillai's Trace .057 2.907b 2.000 96.000 .059

Wilks' Lambda .943 2.907b 2.000 96.000 .059



a. Design: Intercept + grp

Within Subjects Design: site

b. Exact statistic




df Sig. Epsilonb

Greenhouse-Geisser

Huynh-Feldt

Lower-bound

site 0.716 32.021 2 0 0.779 0.798 0.5

Tests the null hypothesis that the error covariance matrix of the orthonormalized transformed dependent variables is proportional to an identity matrix.a. Design: Intercept + grp




Measure: MEASURE_1


site



Huynh-Feldt 102.392 1.595 64.187 84.279 .000

Lower-bound 102.392 1.000 102.392 84.279 .000

site * grp



Huynh-Feldt 4.580 1.595 2.871 3.770 .034

Lower-bound 4.580 1.000 4.580 3.770 .055

Error(site)



Huynh-Feldt 117.846 154.737 .762

Lower-bound 117.846 97.000 1.215

49


Measure: MEASURE_1

Source site Type III Sum of Squares df Mean Square F Sig.

siteLinear 90.284 1 90.284 100.062 .000

Quadratic 12.108 1 12.108 38.728 .000

site * grpLinear 4.548 1 4.548 5.040 .027

Quadratic .032 1 .032 .104 .748

Error(site)Linear 87.521 97 .902

Quadratic 30.325 97 .313


Measure: MEASURE_1



Intercept 4059.473 1 4059.473 5311.083 .000

grp 2.790 1 2.790 3.651 .059

Error 74.141 97 .764

50

glm site1 site2 site3 by grp/wsfactor site 3 difference/contrast(grp)=simple(1).


Measure: MEASURE_1


1 site1

2 site2

3 site3


Value Label N

grp1 Control 51

2 Experimental 48

Multivariate Testsa


site

Pillai's Trace .542 56.803b 2.000 96.000 .000

Wilks' Lambda .458 56.803b 2.000 96.000 .000



site * grp

Pillai's Trace .057 2.907b 2.000 96.000 .059

Wilks' Lambda .943 2.907b 2.000 96.000 .059





b. Exact statistic




df Sig. Epsilonb

Greenhouse-Geisser

Huynh-Feldt

Lower-bound

site 0.716 32.021 2 0 0.779 0.798 0.5




51


Measure: MEASURE_1


site



Huynh-Feldt 102.392 1.595 64.187 84.279 .000

Lower-bound 102.392 1.000 102.392 84.279 .000

site * grp



Huynh-Feldt 4.580 1.595 2.871 3.770 .034

Lower-bound 4.580 1.000 4.580 3.770 .055

Error(site)



Huynh-Feldt 117.846 154.737 .762

Lower-bound 117.846 97.000 1.215


Measure: MEASURE_1


siteLevel 2 vs. Level 1 6.038 1 6.038 8.757 .004

Level 3 vs. Previous 149.060 1 149.060 114.196 .000

site * grpLevel 2 vs. Level 1 2.987 1 2.987 4.333 .040

Level 3 vs. Previous 4.630 1 4.630 3.547 .063

Error(site)Level 2 vs. Level 1 66.873 97 .689

Level 3 vs. Previous 126.614 97 1.305


Measure: MEASURE_1



Intercept 1353.158 1 1353.158 5311.083 .000

grp .930 1 .930 3.651 .059

Error 24.714 97 .255


52


Simple Contrasta Averaged Variable

MEASURE_1

Level 2 vs. Level 1




Std. Error .102

Sig. .059

95% Confidence Interval for DifferenceLower Bound -.008

Upper Bound .395

a. Reference category = 1

Test Results

Measure: MEASURE_1



Contrast .930 1 .930 3.651 .059

Error 24.714 97 .255

glm site1 site2 site3 by grp/wsfactor site 3/print=descriptives/lmatrix "grp2 v 1" grp -1 1/mmatrix "site2 v 1" site1 -1 site2 1 site3 0; "site3 v12" site1 -.5 site2 -.5 site3 1.


Measure: MEASURE_1


1 site1

2 site2

3 site3


Value Label N

grp1 Control 51

2 Experimental 48

53


grp Mean Std. Deviation N

site1

1 Control 4.2935 .81765 51

2 Experimental 4.1695 .79070 48

Total 4.2334 .80301 99

site2

1 Control 3.8727 .58168 51

2 Experimental 4.0962 .75795 48

Total 3.9810 .67878 99

site3

1 Control 2.6391 .77240 51

2 Experimental 3.1216 1.08596 48

Total 2.8730 .96369 99

Multivariate Testsa


site

Pillai's Trace .542 56.803b 2.000 96.000 .000

Wilks' Lambda .458 56.803b 2.000 96.000 .000



site * grp

Pillai's Trace .057 2.907b 2.000 96.000 .059

Wilks' Lambda .943 2.907b 2.000 96.000 .059





b. Exact statistic


Measure: MEASURE_1Within Subjects Effect Mauchly's W Approx. Chi-Square df Sig. Epsilonb

Greenhouse-Geisser


site 0.716 32.021 2 0 0.779 0.798 0.5




54


Measure: MEASURE_1


site



Huynh-Feldt 102.392 1.595 64.187 84.279 .000

Lower-bound 102.392 1.000 102.392 84.279 .000

site * grp



Huynh-Feldt 4.580 1.595 2.871 3.770 .034

Lower-bound 4.580 1.000 4.580 3.770 .055

Error(site)



Huynh-Feldt 117.846 154.737 .762

Lower-bound 117.846 97.000 1.215


Measure: MEASURE_1


siteLinear 90.284 1 90.284 100.062 .000

Quadratic 12.108 1 12.108 38.728 .000

site * grpLinear 4.548 1 4.548 5.040 .027

Quadratic .032 1 .032 .104 .748

Error(site)Linear 87.521 97 .902

Quadratic 30.325 97 .313


Measure: MEASURE_1



Intercept 4059.473 1 4059.473 5311.083 .000

grp 2.790 1 2.790 3.651 .059

Error 74.141 97 .764

55

Custom Hypothesis Tests Index

1

Contrast Coefficients (L' Matrix) LMATRIX Subcommand 1: grp2 v 1

Transformation Coefficients (M Matrix) MMATRIX Subcommand

Contrast Results (K Matrix) Zero Matrix

Custom Hypothesis Tests #1

Contrast Results (K Matrix)a

Contrast Transformed Variable

site2 v 1 site3 v12

L1

Contrast Estimate .348 .433

Hypothesized Value 0 0

Difference (Estimate - Hypothesized) .348 .433

Std. Error .167 .230

Sig. .040 .063


Difference

Lower Bound .016 -.023

Upper Bound .679 .889

a. Based on the user-specified contrast coefficients (L') matrix: grp2 v 1

Multivariate Test Results


Pillai's trace .057 2.907a 2.000 96.000 .059

Wilks' lambda .943 2.907a 2.000 96.000 .059



a. Exact statistic

Univariate Test Results

Source Transformed Variable Sum of Squares df Mean Square F Sig.

Contrastsite2 v 1 2.987 1 2.987 4.333 .040

site3 v12 4.630 1 4.630 3.547 .063

Errorsite2 v 1 66.873 97 .689

site3 v12 126.614 97 1.305

glm test1 test3 with pers1/wsfactor test 2/

56

print=etasq.

General Linear Model


Measure: MEASURE_1


1 test1

2 test3

Multivariate Testsa

Effect Value F Hypothesis df Error df Sig. Partial Eta Squared

test Pillai's Trace 0.071 7.406b 1 97 0.008 0.071

Wilks' Lambda 0.929 7.406b 1 97 0.008 0.071


Roy's Largest Root 0.076 7.406b 1 97 0.008 0.071

test * pers1 Pillai's Trace 0.052 5.330b 1 97 0.023 0.052

Wilks' Lambda 0.948 5.330b 1 97 0.023 0.052


Roy's Largest Root 0.055 5.330b 1 97 0.023 0.052a. Design: Intercept + pers1


b. Exact statistic




df Sig. Epsilonb

Greenhouse-Geisser Huynh-Feldt Lower-bound

test 1 0 0 . 1 1 1

Tests the null hypothesis that the error covariance matrix of the orthonormalized transformed dependent variables is proportional to an identity matrix.a. Design: Intercept + pers1



57


Measure: MEASURE_1Source Type III Sum

of Squaresdf Mean

SquareF Sig. Partial Eta

Squared

test



Huynh-Feldt 4.219 1 4.219 7.406 0.008 0.071

Lower-bound 4.219 1 4.219 7.406 0.008 0.071

test * pers1



Huynh-Feldt 3.036 1 3.036 5.33 0.023 0.052

Lower-bound 3.036 1 3.036 5.33 0.023 0.052

Error(test)

Sphericity Assumed 55.257 97 0.57

Greenhouse-Geisser 55.257 97 0.57

Huynh-Feldt 55.257 97 0.57

Lower-bound 55.257 97 0.57


Measure: MEASURE_1Source test Type III Sum of

Squaresdf Mean Square F Sig. Partial Eta Squared

test Linear 4.219 1 4.219 7.406 0.008 0.071

test * pers1 Linear 3.036 1 3.036 5.33 0.023 0.052

Error(test) Linear 55.257 97 0.57


Measure: MEASURE_1



Squares

df Mean Square F Sig. Partial Eta Squared

Intercept 52.325 1 52.325 73.271 .000 .430

pers1 5.785 1 5.785 8.100 .005 .077

Error 69.270 97 .714

58

compute t31diff=test3 - test1.correlations pers1 t31diff.

Correlations

pers1 t31diff

pers1

Pearson Correlation 1 .228

Sig. (2-tailed) .023

N 99 99

t31diff

Pearson Correlation .228 1


N 99 99

glm t31diff with pers1/print=parameters


Dependent Variable: t31diff


Squares

df Mean Square F Sig.


Intercept 8.438 1 8.438 7.406 .008

pers1 6.072 1 6.072 5.330 .023

Error 110.514 97 1.139

Total 125.993 99



Parameter Estimates




Intercept -1.907 .701 -2.721 .008 -3.297 -.516

pers1 .326 .141 2.309 .023 .046 .607

glm test1 test3 by grp/wsfactor test 2.

59


Measure: MEASURE_1


1 test1

2 test3


Value Label N

grp1 Control 51

2 Experimental 48

Multivariate Testsa


test

Pillai's Trace .074 7.763b 1.000 97.000 .006

Wilks' Lambda .926 7.763b 1.000 97.000 .006



test * grp

Pillai's Trace .029 2.924b 1.000 97.000 .090

Wilks' Lambda .971 2.924b 1.000 97.000 .090





b. Exact statistic



Mauchly's W

Approx. Chi-Square

df Sig. Epsilonb

Greenhouse-Geisser


test 1 0 0 . 1 1 1


60




Measure: MEASURE_1


Squares


test



Huynh-Feldt 4.529 1.000 4.529 7.763 .006

Lower-bound 4.529 1.000 4.529 7.763 .006

test * grp



Huynh-Feldt 1.706 1.000 1.706 2.924 .090

Lower-bound 1.706 1.000 1.706 2.924 .090

Error(test)



Huynh-Feldt 56.588 97.000 .583

Lower-bound 56.588 97.000 .583


Measure: MEASURE_1

Source test Type III Sum of

Squares


test Linear 4.529 1 4.529 7.763 .006

test * grp Linear 1.706 1 1.706 2.924 .090



Measure: MEASURE_1



Squares


Intercept 3945.737 1 3945.737 5542.590 .000

grp 6.001 1 6.001 8.430 .005

61

Error 69.054 97 .712

compute t21diff=test2 - test1.glm t21diff/print=parameters.




Corrected Model .000a 0 . . .

Intercept 6.372 1 6.372 5.500 .021

Error 113.544 98 1.159

Total 119.917 99



Parameter Estimates




Intercept .254 .108 2.345 .021 .039 .468

glm test1 test2 with pers1/wsfactor test 2.


Measure: MEASURE_1


1 test1

2 test2

Multivariate Testsa


test

Pillai's Trace .001 .121b 1.000 97.000 .729

Wilks' Lambda .999 .121b 1.000 97.000 .729

Hotelling's Trace .001 .121b 1.000 97.000 .729

Roy's Largest Root .001 .121b 1.000 97.000 .729

test * pers1 Pillai's Trace .005 .511b 1.000 97.000 .477

Wilks' Lambda .995 .511b 1.000 97.000 .477

62



a. Design: Intercept + pers1


b. Exact statistic



Mauchly's W

Approx. Chi-Square

df

Sig.

Epsilonb

Greenhouse-Geisser

Huynh-Feldt

Lower-

bound

test 1 0 0 . 1 1 1





Measure: MEASURE_1


Squares


test

Sphericity Assumed .071 1 .071 .121 .729

Greenhouse-Geisser .071 1.000 .071 .121 .729

Huynh-Feldt .071 1.000 .071 .121 .729

Lower-bound .071 1.000 .071 .121 .729

test * pers1



Huynh-Feldt .297 1.000 .297 .511 .477

Lower-bound .297 1.000 .297 .511 .477

Error(test)



Huynh-Feldt 56.475 97.000 .582

Lower-bound 56.475 97.000 .582


Measure: MEASURE_1


Squares


test Linear .071 1 .071 .121 .729

test * pers1 Linear .297 1 .297 .511 .477

63



Measure: MEASURE_1



Squares


Intercept 81.396 1 81.396 118.135 .000

pers1 1.459 1 1.459 2.118 .149

Error 66.834 97 .689

glm t21diff with pers1/print=parameters.




Squares


Corrected Model .595a 1 .595 .511 .477

Intercept .141 1 .141 .121 .729

pers1 .595 1 .595 .511 .477

Error 112.950 97 1.164

Total 119.917 99


a. R Squared = .005 (Adjusted R Squared = -.005)

Parameter Estimates




Intercept -.247 .708 -.348 .729 -1.652 1.159

pers1 .102 .143 .715 .477 -.181 .386

temporary.compute pers1 = pers1 - 4.glm test1 test2 with pers1/wsfactor test 2.


64

Measure: MEASURE_1


1 test1

2 test2

Multivariate Testsa


test

Pillai's Trace .009 .926b 1.000 97.000 .338

Wilks' Lambda .991 .926b 1.000 97.000 .338



test * pers1

Pillai's Trace .005 .511b 1.000 97.000 .477

Wilks' Lambda .995 .511b 1.000 97.000 .477





b. Exact statistic



Mauchly's W

Approx. Chi-

Square

df Sig. Epsilonb

Greenhouse-Geisser

Huynh-Feldt

Lower-bound

test 1 0 0 . 1 1 1




65


Measure: MEASURE_1


Squares


test



Huynh-Feldt .539 1.000 .539 .926 .338

Lower-bound .539 1.000 .539 .926 .338

test * pers1



Huynh-Feldt .297 1.000 .297 .511 .477

Lower-bound .297 1.000 .297 .511 .477

Error(test)



Huynh-Feldt 56.475 97.000 .582

Lower-bound 56.475 97.000 .582


Measure: MEASURE_1


Squares


test Linear .539 1 .539 .926 .338

test * pers1 Linear .297 1 .297 .511 .477



Measure: MEASURE_1



Squares


Intercept 1773.060 1 1773.060 2573.363 .000

pers1 1.459 1 1.459 2.118 .149

Error 66.834 97 .689

66

glm test1 test2 with pers1/wsfactor test 2/emmeans=table(test) compare(test) with(pers1=4).


Measure: MEASURE_1


1 test1

2 test2

Multivariate Testsa


test

Pillai's Trace .001 .121b 1.000 97.000 .729

Wilks' Lambda .999 .121b 1.000 97.000 .729



test * pers1

Pillai's Trace .005 .511b 1.000 97.000 .477

Wilks' Lambda .995 .511b 1.000 97.000 .477





b. Exact statistic



Mauchly's W

Approx. Chi-

Square

df Sig. Epsilonb

Greenhouse-Geisser

Huynh-Feldt

Lower-bound

test 1 0 0 . 1 1 1




67

\


Measure: MEASURE_1


Squares


test



Huynh-Feldt .071 1.000 .071 .121 .729

Lower-bound .071 1.000 .071 .121 .729

test * pers1



Huynh-Feldt .297 1.000 .297 .511 .477

Lower-bound .297 1.000 .297 .511 .477

Error(test)



Huynh-Feldt 56.475 97.000 .582

Lower-bound 56.475 97.000 .582


Measure: MEASURE_1


test Linear .071 1 .071 .121 .729

test * pers1 Linear .297 1 .297 .511 .477



Measure: MEASURE_1



Intercept 81.396 1 81.396 118.135 .000

pers1 1.459 1 1.459 2.118 .149

Error 66.834 97 .689

Estimates

Measure: MEASURE_1

test Mean Std. Error 95% Confidence Interval


1 4.559a .124 4.313 4.805

68

2 4.721a .124 4.474 4.968

a. Covariates appearing in the model are evaluated at the following values: pers1 = 4.


Measure: MEASURE_1

(I) test (J) test Mean Difference (I-J) Std. Error Sig.a 95% Confidence Interval for Differencea


11

2 -.162 .168 .338 -.496 .172

21 .162 .168 .338 -.172 .496

2


a. Adjustment for multiple comparisons: Least Significant Difference (equivalent to no adjustments).

Multivariate Tests


Pillai's trace .009 .926a 1.000 97.000 .338

Wilks' lambda .991 .926a 1.000 97.000 .338

Hotelling's trace .010 .926a 1.000 97.000 .338

Roy's largest root .010 .926a 1.000 97.000 .338

Each F tests the multivariate effect of test. These tests are based on the linearly independent pairwise comparisons among


a. Exact statistic

t-test pairs=test3 pers3 with test1 pers1 (paired).

Paired Samples Statistics

Mean N Std. Deviation Std. Error Mean

Pair 1test3 4.31 99 .855 .086

test1 4.61 99 .794 .080

Pair 2pers3 4.71 99 .794 .080

pers1 4.90 99 .763 .077

Paired Samples Correlations

N Correlation Sig.

Pair 1 test3 & test1 99 .126 .214

Pair 2 pers3 & pers1 99 .073 .471

69

Paired Samples Test

Paired Differences t df Sig. (2-tailed)Mean Std.

Deviation

Std. Error Mean

95% Confidence Interval of the

DifferenceLower Upper

Pair 1 test3 - test1 -0.308 1.091 0.11 -0.526 -0.091 -2.812 98 0.006

Pair 2 pers3 - pers1 -0.188 1.06 0.107 -0.399 0.024 -1.762 98 0.081

compute t31diff = test3 - test1.compute p31diff = pers3 - pers1.correlations t31diff p31diff.

Correlations

t31diff p31diff

t31diff

Pearson Correlation 1 -.320


N 99 99

p31diff

Pearson Correlation -.320 1


N 99 99

sort cases by grp.compute swgrp=swgrp+1.if (lag(grp) eq 1 and grp eq 2)swgrp=1.leave swgrp.execute.

select if (swgrp le 20).

varstocases / make resp from site1 site2 site3/index = site(3)//keep = id swgrp grp.

Generated Variables

Name Label

site <none>

resp <none>

Processing Statistics

70

Variables In 41

Variables Out 5

mixed resp by site grp/fixed=intercept site grp site*grp/repeated=site | subject(id) covtype(un)/print=solution testcov.

Model Dimensiona

Number of Levels

Covariance Structure

Number of Parameters

Subject Variables

Number of Subjects

Fixed Effects

Intercept 1 1

site 3 2

grp 2 1

site * grp 6 2

Repeated Effects site 3 Unstructured 6 id 40

Total 15 12 a. Dependent Variable: resp.

Information Criteriaa

-2 Restricted Log Likelihood 280.256

Akaike's Information Criterion

(AIC)292.256

Hurvich and Tsai's Criterion

(AICC)293.041

Bozdogan's Criterion (CAIC) 314.673

Schwarz's Bayesian Criterion (BIC) 308.673

The information criteria are displayed in smaller-

is-better forms.

a. Dependent Variable: resp.

Type III Tests of Fixed Effectsa

Source Numerator df Denominator df F Sig.

Intercept 1 38.000 2072.696 .000

site 2 38.000 32.193 .000

grp 1 38.000 .075 .786

site * grp 2 38.000 .495 .613


71

Estimates of Fixed Effectsa

Parameter Estimate Std. Error df t Sig. 95% Confidence Interval

Fraction

Missing

Info.

Relative Increase Variance

Relative Efficiency

Lower Bound

Upper Bound

Intercept 2.771982 0.216238 38 12.819 0 2.334231 3.209734

[site=1] 1.452649 0.299594 38 4.849 0 0.846154 2.059145

[site=2] 1.273138 0.252043 38 5.051 0 0.762904 1.783371

[site=3] 0b 0 . . . . .

[grp=1] -0.213282 0.305807 38 -0.697 0.49 -0.832356 0.405792

[grp=2] 0b 0 . . . . .

[site=1] * [grp=1] 0.364135 0.42369 38 0.859 0.395 -0.49358 1.221849

[site=1] * [grp=2] 0b 0 . . . . .

[site=2] * [grp=1] 0.144 0.356442 38 0.404 0.688 -0.577579 0.865579

[site=2] * [grp=2] 0b 0 . . . . .

[site=3] * [grp=1] 0b 0 . . . . .

[site=3] * [grp=2] 0b 0 . . . . .


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

Covariance ParametersEstimates of Covariance Parametersa

Parameter Estimate Std. Erro

r

Wald Z

Sig. 95% Confidence Interval

Fraction Missing

Info.

Relative Increase Varianc

e

Relative Efficiency

Lower Bound

Upper Bound

Repeated Measures

UN (1,1) 0.596764 0.14 4.359 0 0.380648 0.935582

UN (2,1) 0.225965 0.09 2.42 0.016 0.04293 0.408999

UN (2,2) 0.469767 0.11 4.359 0 0.299643 0.736482

UN (3,1) -0.13159 0.12-

1.069 0.285 -0.37277 0.109587

UN (3,2) 0.067219 0.11 0.622 0.534 -0.1446 0.279039

UN (3,3) 0.935179 0.21 4.359 0 0.596507 1.466135


Rangkuman

Sintaksis yang dikumpulkan dari tulisan Alan Taylor merupakan contoh pemakaian sintaksis.

Sintaksis seperti disajikan di atas tidak terdapat dalam buku-buku SPSS yang telah

72

diterbitkan di Indonesia dan para penerbit mungkin enggan menerbitkan buku sintaksis yang

dapat diaplikasikan karena kemungkinan penjualan buku tersebut akan sangat rendah.

Usaha memperkenalkan sintaksis dalam IBM SPSS Statistics akan terus dilakukan sejalan

dengan pemakaian sintaksis IBM SPSS Statistics di berbagai negara lain. Banyak sumber

bacaan tentang IBM SPSS Statistics yang mengandung pembahasan mengenai sintaksis.

Penulis mengharap kritik atas isi tulisan ini.

Permata Depok Regency, 18 April 2015

73

Alan Taylor

Documents

Transcript of Alan Taylor