You are on page 1of 11

John E.

Wick
LAB C1
EDUC 746

LAB C1 - MANOVA

The purpose of this study was to determine if the treatment administered affects

the scores on the WRAT-R and WRAT-A assessments for children with disabilities.

There are three research questions that are being examined through analysis of the data

collected:

• Does the treatment have a main effect on the WRAT-R and WRAT-A
scores for learning disabled children?

• Does the level of disability have a main effect upon the WRAT-R and
WRAT-A for learning disabled children?

• Does the treatment effect depend upon the level of disability (Is there an
interaction)?

There are four variables being examined in this study, two independent variables

[IV] and two dependent variables [DV]. The first IV is Group. This is a categorical

variable, which has been dummy coded as 1=treatment received, 2=no treatment

received. The next IV is also categorical; it is Degree of Disability and has been dummy

coded as 1=mild, 2=moderate, 3=severe.

The participants for this study were learning disabled children with a sample size

of n=18.

Multivariate Analysis of Variance [MANOVA] was performed to analyze the data

collected. Results of the MANOVA revealed that the treatment does have a significant

main effect upon the WRAT-R and WRAT-A assessments, Wilks’ Λ = .138, F (2,11) =

34.436, p < .05, η2p = .862. Furthermore, the treatment effect is significant on the WRAT-
John E. Wick
LAB C1
EDUC 746
R, F (1,12) = 46.123, p < .05, η2p = .794. The treatment effect is also significant for the

WRAT-A, F (1,12) = 33.246, p <.05, η2p = .735.

On the WRAT-R, the treatment demonstrated the greatest increase for the mild

degree of disability (M=106.6667, SD=8.50490), followed by the moderate degree of

disability (M=100.00, SD=5.00), and then severe disability level (M=93.00,

SD=6.08276).

On the WRAT-A, the treatment demonstrated the greatest increase for the mild

degree of disability (M=103.6667, SD=5.13160), followed by the moderate degree of

disability (M=99.3333, SD=5.13160), and then the sever disability (M=86.00,

SD=8.54400).

Results from the MANOVA analysis also indicates that the level of disability

does have a main effect upon the WRAT-R and WRAT-A for learning disabled children,

Wilks’ Λ = .255, F (4,22) = 5.386, p < .05, η2p = .495. The main effect of the level of

disability upon the WRAT-R was shown to be significant, F (2,12) = 5.744, p < .05, η2p

= .489. The level of difficulty effect was also found to be significant for the WRAT-A, F

(2,12) = 12.535, p < .05, η2p = .676.

On the WRAT-R, the level of learning disability that demonstrated the greatest

increase within the treatment group was the mild degree of disability (M=106.6667,

SD=8.50490), followed by the moderate degree of disability (M=100.00, SD=5.00), and

then severe disability level (M=93.00, SD=6.08276).

On the WRAT-A, the level of learning disability that demonstrated the greatest

increase within the treatment goup was the mild degree of disability (M=103.6667,

SD=5.13160), followed by the moderate degree of disability (M=99.3333, SD=5.13160),


John E. Wick
LAB C1
EDUC 746
and then the sever disability (M=86.00, SD=8.54400).

Results from the MANOVA analysis also indicated that the treatment effect does

not depend upon the level of disability (there is no interaction), Wilks’ Λ = .908, F (4,22)

= .272, p > .05, η2p = .047.


John E. Wick
LAB C1
EDUC 746
NEW FILE.
DATASET NAME DataSet1 WINDOW=FRONT.
SAVE OUTFILE='/Users/John/Documents/APU/EdD/ADV SATS/LAB C/LABC1
Wick DATA.sav'
/COMPRESSED.
GLM WratR WratA BY Groups Degree
/METHOD=SSTYPE(3)
/INTERCEPT=INCLUDE
/PRINT=DESCRIPTIVE ETASQ TEST(SSCP) RSSCP
/CRITERIA=ALPHA(.05)
/DESIGN= Groups Degree Groups*Degree.

General Linear Model


[DataSet1] /Users/John/Documents/APU/EdD/ADV SATS/LAB C/LABC1
Wick DATA.sav

Between-Subjects Factors
Value Label N
Groups 1.00 Treatment 9
2.00 Control 9
Degree 1.00 Mild 6
2.00 Moderate 6
3.00 Severe 6
John E. Wick
LAB C1
EDUC 746

Descriptive Statistics
Groups Degree Mean Std. Deviation N
WratR Treatment Mild 106.6667 8.50490 3
Moderate 100.0000 5.00000 3
Severe 93.0000 6.08276 3
Total 99.8889 8.28318 9
Control Mild 85.0000 5.00000 3
Moderate 77.6667 7.50555 3
Severe 72.3333 7.50555 3
Total 78.3333 8.04674 9
Total Mild 95.8333 13.40771 6
Moderate 88.8333 13.49691 6
Severe 82.6667 12.86338 6
Total 89.1111 13.62907 18
WratA Treatment Mild 103.6667 5.13160 3
Moderate 99.3333 5.13160 3
Severe 86.0000 8.54400 3
Total 96.3333 9.74679 9
Control Mild 89.3333 7.37111 3
Moderate 76.6667 7.57188 3
Severe 68.3333 5.68624 3
Total 78.1111 10.94811 9
Total Mild 96.5000 9.69020 6
Moderate 88.0000 13.69671 6
Severe 77.1667 11.65190 6
Total 87.2222 13.74796 18

Bartlett's Test of Sphericitya


Likelihood Ratio .971
Approx. Chi-Square .036
df 2
Sig. .982
Tests the null hypothesis that the
residual covariance matrix is
proportional to an identity matrix.
John E. Wick
LAB C1
EDUC 746
Bartlett's Test of Sphericitya
Likelihood Ratio .971
Approx. Chi-Square .036
df 2
Sig. .982
Tests the null hypothesis that the
residual covariance matrix is
proportional to an identity matrix.
a. Design: Intercept + Groups +
Degree + Groups * Degree

c
Multivariate Tests
Partial Eta
Effect Value F Hypothesis df Error df Sig. Squared
a
Intercept Pillai's Trace .998 2687.779 2.000 11.000 .000 .998
a
Wilks' Lambda .002 2687.779 2.000 11.000 .000 .998
a
Hotelling's Trace 488.687 2687.779 2.000 11.000 .000 .998
a
Roy's Largest Root 488.687 2687.779 2.000 11.000 .000 .998
a
Groups Pillai's Trace .862 34.436 2.000 11.000 .000 .862
a
Wilks' Lambda .138 34.436 2.000 11.000 .000 .862
a
Hotelling's Trace 6.261 34.436 2.000 11.000 .000 .862
a
Roy's Largest Root 6.261 34.436 2.000 11.000 .000 .862
Degree Pillai's Trace .750 3.604 4.000 24.000 .019 .375
a
Wilks' Lambda .255 5.386 4.000 22.000 .004 .495
Hotelling's Trace 2.895 7.238 4.000 20.000 .001 .591
b
Roy's Largest Root 2.887 17.323 2.000 12.000 .000 .743
Groups * Degree Pillai's Trace .092 .290 4.000 24.000 .882 .046
a
Wilks' Lambda .908 .272 4.000 22.000 .893 .047
Hotelling's Trace .101 .252 4.000 20.000 .905 .048
b
Roy's Largest Root .098 .588 2.000 12.000 .571 .089
a. Exact statistic
b. The statistic is an upper bound on F that yields a lower bound on the significance level.
c. Design: Intercept + Groups + Degree + Groups * Degree
John E. Wick
LAB C1
EDUC 746
Tests of Between-Subjects Effects

Type III Sum of Partial Eta


Source Dependent Variable Squares df Mean Square F Sig. Squared
a
Corrected Model WratR 2613.778 5 522.756 11.531 .000 .828
b
WratA 2673.778 5 534.756 11.898 .000 .832
Intercept WratR 142934.222 1 142934.222 3152.961 .000 .996
WratA 136938.889 1 136938.889 3046.848 .000 .996
Groups WratR 2090.889 1 2090.889 46.123 .000 .794
WratA 1494.222 1 1494.222 33.246 .000 .735
Degree WratR 520.778 2 260.389 5.744 .018 .489
WratA 1126.778 2 563.389 12.535 .001 .676
Groups * Degree WratR 2.111 2 1.056 .023 .977 .004
WratA 52.778 2 26.389 .587 .571 .089
Error WratR 544.000 12 45.333
WratA 539.333 12 44.944
Total WratR 146092.000 18
WratA 140152.000 18
Corrected Total WratR 3157.778 17

WratA 3213.111 17

a. R Squared = .828 (Adjusted R Squared = .756)


b. R Squared = .832 (Adjusted R Squared = .762)

Between-Subjects SSCP Matrix


WratR WratA
Hypothesis Intercept WratR 142934.222 139904.444
WratA 139904.444 136938.889
Groups WratR 2090.889 1767.556
WratA 1767.556 1494.222
Degree WratR 520.778 761.722
WratA 761.722 1126.778
Groups * Degree WratR 2.111 5.278
WratA 5.278 52.778
Error WratR 544.000 31.000
WratA 31.000 539.333
Based on Type III Sum of Squares
John E. Wick
LAB C1
EDUC 746

Residual SSCP Matrix


WratR WratA
Sum-of-Squares and WratR 544.000 31.000
Cross-Products WratA 31.000 539.333
Covariance WratR 45.333 2.583
WratA 2.583 44.944
Correlation WratR 1.000 .057
WratA .057 1.000
Based on Type III Sum of Squares
John E. Wick
LAB C1
EDUC 746

MATRIX.
compute sscpgp={2090.889, 1767.556; 1767.556, 1494.222}.
print sscpgp.
compute sscpdg={520.778, 761.722; 761.722, 1126.778}.
print sscpdg.
compute sscpgd={2.111, 5.278; 5.278, 52.778}.
print sscpgd.
compute sscpwg={544, 31; 31, 539.333}.
print sscpwg.
compute sscptgp=sscpgp+sscpwg.
print sscptgp.
compute dettgp=det(sscptgp).
print dettgp.
compute sscptdg=sscpdg+sscpwg.
print sscptdg.
compute dettdg=det(sscptdg).
print dettdg.
compute sscptgd=sscpgd+sscpwg.
print sscptgd.
compute dettgd=det(sscptgd).
print dettgd.
compute detwg=det(sscpwg).
print detwg.
compute lambdagp=detwg/dettgp.
print lambdagp.
compute lambdadg=detwg/dettdg.
print lambdadg.
compute lambdagd=detwg/dettgd.
print lambdagd.
end matrix.
John E. Wick
LAB C1
EDUC 746
Matrix
[DataSet1] /Users/John/Documents/APU/EdD/ADV SATS/LAB C/LABC1 Wick DATA.sav

Run MATRIX procedure:

SSCPGP
2090.889000 1767.556000
1767.556000 1494.222000

SSCPDG
520.778000 761.722000
761.722000 1126.778000

SSCPGD
2.11100000 5.27800000
5.27800000 52.77800000

SSCPWG
544.0000000 31.0000000
31.0000000 539.3330000

SSCPTGP
2634.889000 1798.556000
1798.556000 2033.555000

DETTGP
10 ** 6 X
2.123388015

SSCPTDG
1064.778000 792.722000
792.722000 1666.111000

DETTDG
10 ** 6 X
1.145630169

SSCPTGD
546.1110000 36.2780000
36.2780000 592.1110000

DETTGD
10 ** 5 X
3.220422370

DETWG
10 ** 5 X
2.924361520

LAMBDAGP
.1377214856

LAMBDADG
.2552622652

LAMBDAGD
.9080676954

------ END MATRIX -----


John E. Wick
LAB C1
EDUC 746

You might also like