MANOVA - academic.csuohio.edu · MANOVA Tracy Markis COM 531 Spring 2009 The effect of color vs....

MANOVA

Tracy Markis COM 531 Spring 2009

The effect of color vs. black-and-white expectation and viewing condition on subsequent cognitive processing, specifically attention and recall, was examined using Multivariate Analysis of Variance (MANOVA). Dependent Variables Q2. To what extent did you feel mentally immersed in the film? Q24. I paid more attention to the film environment than I did to my own thoughts V125. Recallf V126. Recallv Independent Variables Q108. Condition the participant experienced Q59. Expect3

MANOVA MODEL

Condition(2) Expect3(3) Condition x Expect3

DV1 – To what extent did you feel mentally immersed

in the film? DV2 – I paid more attention to the film environment

than I did to my own thoughts DV3 – Recallf DV4 - Recallv

1

Variable Details

Dependent Variables Q2. To what extent did you feel mentally immersed in the film? Not at all 1 2 3 4 5 6 7 Very much Q24. I paid more attention to the film environment than I did to my own thoughts Strongly 1 2 3 4 5 6 7 Strongly Disagree Agree V125. Recallf = mean(refugees,cities,camps,roosevel,godbless)

FACTUAL – Recall - yes = 2/no = 0 How many refugees were trekking through Europe? Name two of the German cities that the allies bombed heavily. Name the two Nazi concentration camps shown in this documentary. On what date did FDR die? What was the nationality of the women who learned and sang "God Bless America?"

V126. Recallv = mean(refugees,cities,camps,roosevel,godbless)

VISUAL– Recall - yes = 2/no = 0 How many men were dancing to the Yankee Doodlers? Describe the uniforms worn by concentration camp prisoners. What was the second concentration camp quarantined for? The first USO dancer wore what type of skirt? How many paratroopers are seen parachuting from an allied plane?

Independent Variables Q108. Condition the participant experienced (1=color footage; 2=black and white footage) Q59. I expected the film to be in black-and-white RECODED Q59 into a categorical variable named Expect3. Strongly 1 2 3 4 5 6 7 Strongly

Disagree Agree EXPECT3 1 2 3 Do not expect Don’t know Expect B &W B & W

2

ASSUMPTIONS IN MANOVA

Dependent variables ought to correlate with each other. Correlations

** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed).

To what extent did you feel

mentally immersed in

the film?

I paid more attention to the

film environment

than I did to my own thoughts.

COMPUTE recallf =

mean(refugees,cities,camps,roosevel,god

bless) (COMPUTE)

COMPUTE recallv =

mean(yankdood,outfits,quaranti,usoskirt,p

artroop) (COMPUTE)

Pearson Correlation 1 .426(**) .215(*) .126Sig. (2-tailed) .000 .025 .191

To what extent did you feel mentally immersed in the film?

N 109 109 109 109Pearson Correlation .426(**) 1 .237(*) .012Sig. (2-tailed) .000 .013 .903

I paid more attention to the film environment than I did to my own thoughts.

N 109 109 109 109Pearson Correlation .215(*) .237(*) 1 .362(**)Sig. (2-tailed) .025 .013 .000

COMPUTE recallf = mean(refugees,cities,camps,roosevel,godbless) (COMPUTE) N

109 109 109 109

Pearson Correlation .126 .012 .362(**) 1Sig. (2-tailed) .191 .903 .000

COMPUTE recallv = mean(yankdood,outfits,quaranti,usoskirt,partroop) (COMPUTE) N 109 109 109 109

3

Assumes independence of observations (theoretical – can’t really test) – lack of independence among observations means that the responses in each cell (group) are not made independently of responses in any other group (Hair et al., 2006).

Assumes equality of the variance/covariance matrix (of the DV’s and across groups) Box's Test of Equality of Covariance Matrices(a)

Tests the null hypothesis that the observed covariance matrices of the dependent variables are equal across groups. a Design: Intercept+condtion+expect3+condtion * expect3 Equality of the variance/covariance matrices is covered through Box’s M. Box’s M is not significant (p = .536), indicating that this assumption is satisfied.

Assumes linearity among the DVs – can check scatterplots.

MANOVA is sensitive to outliers – check for them.

Assumes univariate and multivariate normality of the DVs. Visual inspection of frequency distributions for each dependent variable.

Box's M 57.712 F .969 df1 50 df2 3657.641 Sig. .536

4


Frequency Percent Valid Percent Cumulative

Percent Not at all 3 2.8 2.8 2.82.00 6 5.5 5.5 8.33.00 11 10.1 10.1 18.34.00 28 25.7 25.7 44.05.00 30 27.5 27.5 71.66.00 18 16.5 16.5 88.1Very Much 13 11.9 11.9 100.0

Valid

Total 109 100.0 100.0


8.006.004.002.000.00

Freq

uenc

y

30

20

10

0


Mean =4.67�Std. Dev. =1.466�

N =109

5



Percent Strongly Disagree 14 12.8 12.8 12.8 2.00 12 11.0 11.0 23.9 3.00 7 6.4 6.4 30.3 4.00 15 13.8 13.8 44.0 5.00 20 18.3 18.3 62.4 6.00 19 17.4 17.4 79.8 Strongly Agree 22 20.2 20.2 100.0

Valid

Total 109 100.0 100.0


8.006.004.002.000.00

Freq

uenc

y

25

20

15

10

5

0


Mean =4.47�Std. Dev. =2.044�

N =109

6

COMPUTE recallf = mean(refugees,cities,camps,roosevel,godbless) (COMPUTE)


Percent .00 54 49.5 49.5 49.5.20 8 7.3 7.3 56.9.40 19 17.4 17.4 74.3.60 8 7.3 7.3 81.7.80 12 11.0 11.0 92.71.00 1 .9 .9 93.61.20 5 4.6 4.6 98.21.40 2 1.8 1.8 100.0

Valid

Total 109 100.0 100.0


1.501.000.500.00-0.50

Freq

uenc

y

60

50

40

30

20

10

0


Mean =0.31�Std. Dev. =0.383�

N =109

7

COMPUTE recallv = mean(yankdood,outfits,quaranti,usoskirt,partroop) (COMPUTE)


Percent .00 24 22.0 22.0 22.0.40 18 16.5 16.5 38.5.80 30 27.5 27.5 66.11.20 18 16.5 16.5 82.61.60 13 11.9 11.9 94.52.00 6 5.5 5.5 100.0

Valid

Total 109 100.0 100.0


2.502.001.501.000.500.00-0.50

Freq

uenc

y

30

20

10

0


Mean =0.79�Std. Dev. =0.594�

N =109

8

MANOVA STEPS IN SPSS

Analyze → General Linear Model → Multivariate Enter the dependent variables into Dependent Variables box Enter the independent variables into the Fixed Factor(s) box

9

Click Model → Full Factorial (SPSS default) Click Continue

10

Click Plots Multivariate: Profile Plots Horizontal Axis: condition Separate Lines: expect3

Click Add after each plot Horizontal Axis: expect3 Separate Lines: condition Click Add and second plot is added

11

Click Post Hoc I chose Tukey’s and Scheffe – you can choose whichever one’s you like. Run post hoc tests for Expect3

13

Select Options Display Means for: put in all factors and factor interactions Under Display check: Descriptive statistics Estimates of effect size Observed power Homogeneity tests Click Continue

14

Click OK to run MANOVA

15

GET FILE='C:\Documents and Settings\1002678\Desktop\DennyManova.sav'.GLM Q2mentimme Q24paidatt recallf recallv BY condtion expect3 /METHOD = SSTYPE(3) /INTERCEPT = INCLUDE /POSTHOC = expect3 ( TUKEY SCHEFFE ) /PLOT = PROFILE( condtion*expect3 expect3*condtion ) /EMMEANS = TABLES(condtion) /EMMEANS = TABLES(expect3) /EMMEANS = TABLES(condtion*expect3) /PRINT = DESCRIPTIVE ETASQ OPOWER HOMOGENEITY /CRITERIA = ALPHA(.05) /DESIGN = condtion expect3 condtion*expect3 .

General Linear Model

Between-Subjects Factors

Colorfootage 48

Black andwhitefootage

61

1-Do notexpect B&W 28

2-DK 163-ExpectB&W 65

1

2

Condition the participantexperienced.

1.00

2.003.00

expect3

Value Label N

Page 1

16

Descriptive Statistics

4.8824 1.691245.2857 .951194.9583 1.160184.9792 1.328734.1818 1.537414.0000 1.414214.5854 1.564864.4262 1.532534.6071 1.640674.5625 1.364734.7231 1.430774.6697 1.466003.7059 2.201275.1429 1.772815.0833 2.041244.6042 2.131223.8182 1.940014.0000 1.870834.5854 2.024544.3607 1.983543.7500 2.066044.5000 1.861904.7692 2.029234.4679 2.04382

.3176 .34683

.7143 .38048

.3250 .36742

.3792 .38091

.1636 .23355

.0889 .14530

.3073 .42918

.2492 .37710

.2571 .31202

.3625 .41453

.3138 .40461

.3064 .38255

.8471 .646261.3143 .55205

.7500 .58160

.8667 .61931

.4727 .66497

.4444 .46667

.8488 .53062

.7213 .56954

.7000 .66778

.8250 .66081

.8123 .54758

.7853 .59361

expect31-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal

Condition the participanti dColor footage

Black and white footage

Total

Color footage


Total

Color footage


Total

Color footage


Total

To what extent did youfeel mentally immersed inthe film?

I paid more attention tothe film environment than Idid to my own thoughts.

COMPUTE recallf =mean(refugees,cities,camps,roosevel,godbless)(COMPUTE)

COMPUTE recallv =mean(yankdood,outfits,quaranti,usoskirt,partroop)(COMPUTE)

Mean Std. Deviation

Page 2

17

Descriptive Statistics

177

244811

94161281665

10917

7244811

94161281665

10917

7244811

94161281665

10917

7244811

94161281665

109

expect31-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal1-Do not expect B&W2-DK3-Expect B&WTotal

Condition the participanti dColor footage


Total

Color footage


Total

Color footage


Total

Color footage


Total

To what extent did youfeel mentally immersed inthe film?




N

Page 3

18

Box's Test of Equality of Covariance Matricesa

57.712.969

503657.641

.536

Box's MFdf1df2Sig.

Tests the null hypothesis that the observed covariance matrices of the dependent variables are equal across groups.Design: Intercept+condtion+expect3+condtion * expect3a.

Multivariate Testsd

.903 231.536b 4.000 100.000 .000

.097 231.536b 4.000 100.000 .0009.261 231.536b 4.000 100.000 .0009.261 231.536b 4.000 100.000 .000

.147 4.294b 4.000 100.000 .003

.853 4.294b 4.000 100.000 .003

.172 4.294b 4.000 100.000 .003

.172 4.294b 4.000 100.000 .003

.075 .978 8.000 202.000 .454

.926 .978b 8.000 200.000 .455

.079 .977 8.000 198.000 .455

.067 1.704c 4.000 101.000 .155

.137 1.858 8.000 202.000 .068

.865 1.887b 8.000 200.000 .064

.155 1.915 8.000 198.000 .060

.141 3.565c 4.000 101.000 .009

Pillai's TraceWilks' LambdaHotelling's TraceRoy's Largest RootPillai's TraceWilks' LambdaHotelling's TraceRoy's Largest RootPillai's TraceWilks' LambdaHotelling's TraceRoy's Largest RootPillai's TraceWilks' LambdaHotelling's TraceRoy's Largest Root

EffectIntercept

condtion

expect3

condtion * expect3

Value F Hypothesis df Error df Sig.

Page 4

19

Multivariate Testsd

.903 926.146 1.000

.903 926.146 1.000

.903 926.146 1.000

.903 926.146 1.000

.147 17.175 .919

.147 17.175 .919

.147 17.175 .919

.147 17.175 .919

.037 7.822 .449

.038 7.820 .449

.038 7.816 .448

.063 6.815 .506

.069 14.862 .776

.070 15.093 .784

.072 15.316 .791

.124 14.259 .855

Pillai's TraceWilks' LambdaHotelling's TraceRoy's Largest RootPillai's TraceWilks' LambdaHotelling's TraceRoy's Largest RootPillai's TraceWilks' LambdaHotelling's TraceRoy's Largest RootPillai's TraceWilks' LambdaHotelling's TraceRoy's Largest Root

EffectIntercept

condtion

expect3

condtion * expect3

Partial EtaSquared

Noncent.Parameter

ObservedPowera

Computed using alpha = .05a. Exact statisticb. The statistic is an upper bound on F that yields a lower bound on the significance level.c. Design: Intercept+condtion+expect3+condtion * expect3d.

Levene's Test of Equality of Error Variancesa

1.534 5 103 .186

.581 5 103 .715

2.741 5 103 .023

.415 5 103 .838

To what extent did youfeel mentally immersed inthe film?I paid more attention tothe film environment than Idid to my own thoughts.COMPUTE recallf =mean(refugees,cities,camps,roosevel,godbless)(COMPUTE)COMPUTE recallv =mean(yankdood,outfits,quaranti,usoskirt,partroop)(COMPUTE)

F df1 df2 Sig.

Tests the null hypothesis that the error variance of the dependent variable is equal across groups.Design: Intercept+condtion+expect3+condtion * expect3a.

Page 5

20

Tests of Between-Subjects Effects

12.371b

5 2.474 1.160

29.330c

5 5.866 1.432

1.825d

5 .365 2.689

4.339e

5 .868 2.651

1656.283 1 1656.283 776.362

1476.431 1 1476.431 360.526

7.821 1 7.821 57.622

46.571 1 46.571 142.265

11.848 1 11.848 5.554

4.974 1 4.974 1.214

1.353 1 1.353 9.965

2.793 1 2.793 8.531

1.107 2 .553 .259

21.372 2 10.686 2.609

.263 2 .131 .968

.562 2 .281 .858

Dependent VariableTo what extent did youfeel mentally immersed inthe film?I paid more attention tothe film environment than Idid to my own thoughts.COMPUTE recallf =mean(refugees,cities,camps,roosevel,godbless)(COMPUTE)COMPUTE recallv =mean(yankdood,outfits,quaranti,usoskirt,partroop)(COMPUTE)To what extent did youfeel mentally immersed inthe film?I paid more attention tothe film environment than Idid to my own thoughts.COMPUTE recallf =mean(refugees,cities,camps,roosevel,godbless)(COMPUTE)COMPUTE recallv =mean(yankdood,outfits,quaranti,usoskirt,partroop)(COMPUTE)To what extent did youfeel mentally immersed inthe film?I paid more attention tothe film environment than Idid to my own thoughts.COMPUTE recallf =mean(refugees,cities,camps,roosevel,godbless)(COMPUTE)COMPUTE recallv =mean(yankdood,outfits,quaranti,usoskirt,partroop)(COMPUTE)To what extent did youfeel mentally immersed inthe film?I paid more attention tothe film environment than Idid to my own thoughts.COMPUTE recallf =mean(refugees,cities,camps,roosevel,godbless)(COMPUTE)COMPUTE recallv =mean(yankdood,outfits,quaranti,usoskirt,partroop)(COMPUTE)

SourceCorrected Model

Intercept

condtion

expect3

Type III Sumof Squares df Mean Square F

Page 6

21


2.699 2 1.350 .633

4.033 2 2.017 .492

1.155 2 .577 4.253

3.301 2 1.650 5.042

219.739 103 2.133

421.807 103 4.095

13.980 103 .136

33.717 103 .327

2609.000 109

2627.000 109

26.040 109

105.280 109

232.110 108

451.138 108

15.806 108

38.057 108


Sourcecondtion * expect3

Error

Total

Corrected Total

Type III Sumof Squares df Mean Square F

Page 7

22


.334 .053 5.799 .398

.219 .065 7.162 .487

.025 .115 13.446 .797

.027 .114 13.255 .790

.000 .883 776.362 1.000

.000 .778 360.526 1.000

.000 .359 57.622 1.000

.000 .580 142.265 1.000

.020 .051 5.554 .646

.273 .012 1.214 .194

.002 .088 9.965 .878

.004 .076 8.531 .825

.772 .005 .519 .090

.078 .048 5.219 .509

.383 .018 1.937 .214

.427 .016 1.715 .194


SourceCorrected Model

Intercept

condtion

expect3

Sig.Partial EtaSquared

Noncent.Parameter

ObservedPowera

Page 8

23


.533 .012 1.265 .153

.613 .009 .985 .129

.017 .076 8.507 .732

.008 .089 10.084 .807


Sourcecondtion * expect3

Error

Total

Corrected Total

Sig.Partial EtaSquared

Noncent.Parameter

ObservedPowera

Computed using alpha = .05a. R Squared = .053 (Adjusted R Squared = .007)b. R Squared = .065 (Adjusted R Squared = .020)c.

Page 9

24


R Squared = .115 (Adjusted R Squared = .073)d. R Squared = .114 (Adjusted R Squared = .071)e.

Page 10

25

Estimated Marginal Means

1. Condition the participant experienced.

5.042 .240 4.566 5.518

4.256 .232 3.796 4.7154.644 .333 3.984 5.304

4.135 .321 3.498 4.771

.452 .061 .332 .572

.187 .058 .071 .303

.970 .094 .784 1.157

.589 .091 .409 .769

Condition the participantexperienced.Color footage


Color footageBlack and white footage

Color footageBlack and white footageColor footageBlack and white footage

Dependent VariableTo what extent did youfeel mentally immersed inthe film?I paid more attention tothe film environment than Idid to my own thoughts.

COMPUTE recallf =mean(refugees,cities,camps roosevel godbless)COMPUTE recallv =mean(yankdood,outfits,quaranti usoskirt partroop)

Mean Std. Error Lower Bound Upper Bound95% Confidence Interval

2. expect3

4.532 .283 3.972 5.0934.643 .368 3.913 5.3734.772 .188 4.400 5.1443.762 .392 2.986 4.5394.571 .510 3.560 5.583

4.834 .260 4.319 5.350.241 .071 .099 .382.402 .093 .217 .586

.316 .047 .222 .410

.660 .111 .440 .879

.879 .144 .593 1.165

.799 .074 .654 .945

expect31-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK

3-Expect B&W

1-Do not expect B&W2-DK3-Expect B&W

1-Do not expect B&W2-DK3-Expect B&W

Dependent VariableTo what extent did youfeel mentally immersed inthe film?




Mean Std. Error Lower Bound Upper Bound95% Confidence Interval

Page 11

26

3. Condition the participant experienced. * expect3

4.882 .3545.286 .5524.958 .2984.182 .4404.000 .4874.585 .2283.706 .4915.143 .7655.083 .4133.818 .6104.000 .6754.585 .316

.318 .089

.714 .139

.325 .075

.164 .111

.089 .123

.307 .058

.847 .1391.314 .216

.750 .117

.473 .173

.444 .191

.849 .089

expect31-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W



Color footage


Color footage


Color footage






Mean Std. Error

Page 12

27

3. Condition the participant experienced. * expect3

4.180 5.5854.191 6.3814.367 5.5503.308 5.0553.034 4.9664.133 5.0382.732 4.6793.626 6.6604.264 5.9032.608 5.0282.662 5.3383.959 5.212

.140 .495

.438 .990

.176 .474-.057 .384-.155 .332.193 .421.572 1.122.885 1.743.518 .982.131 .815.066 .823.672 1.026

expect31-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W1-Do not expect B&W2-DK3-Expect B&W



Color footage


Color footage


Color footage






Lower Bound Upper Bound95% Confidence Interval

Post Hoc Tests

expect3

Page 13

28

Multiple Comparisons

.0446 .45774-.1159 .33017-.0446 .45774-.1606 .40763.1159 .33017.1606 .40763.0446 .45774

-.1159 .33017-.0446 .45774-.1606 .40763.1159 .33017.1606 .40763

-.7500 .63420-1.0192 .45745

.7500 .63420-.2692 .564761.0192 .45745

.2692 .56476-.7500 .63420

-1.0192 .45745.7500 .63420

-.2692 .564761.0192 .45745

.2692 .56476-.1054 .11546-.0567 .08328.1054 .11546.0487 .10282.0567 .08328

-.0487 .10282-.1054 .11546-.0567 .08328.1054 .11546.0487 .10282.0567 .08328

-.0487 .10282-.1250 .17931-.1123 .12933.1250 .17931.0127 .15967.1123 .12933

-.0127 .15967-.1250 .17931-.1123 .12933.1250 .17931.0127 .15967.1123 .12933

-.0127 .15967

(J) expect32-DK3-Expect B&W1-Do not expect B&W3-Expect B&W1-Do not expect B&W2-DK2-DK3-Expect B&W1-Do not expect B&W3-Expect B&W1-Do not expect B&W2-DK2-DK3-Expect B&W1-Do not expect B&W3-Expect B&W1-Do not expect B&W2-DK2-DK3-Expect B&W1-Do not expect B&W3-Expect B&W1-Do not expect B&W2-DK2-DK3-Expect B&W1-Do not expect B&W3-Expect B&W1-Do not expect B&W2-DK2-DK3-Expect B&W1-Do not expect B&W3-Expect B&W1-Do not expect B&W2-DK2-DK3-Expect B&W1-Do not expect B&W3-Expect B&W1-Do not expect B&W2-DK2-DK3-Expect B&W1-Do not expect B&W3-Expect B&W1-Do not expect B&W2-DK

(I) expect31-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

Tukey HSD

Scheffe

Tukey HSD

Scheffe

Tukey HSD

Scheffe

Tukey HSD

Scheffe





MeanDifference

(I-J) Std. Error

Based on observed means.

Page 14

29


.995

.934

.995

.918

.934

.918

.995

.940

.995

.925

.940

.925

.466

.071

.466

.882

.071

.882

.499

.089

.499

.893

.089

.893

.634

.775

.634

.884

.775

.884

.661

.794

.661

.894

.794

.894

.766

.661

.766

.997

.661

.997

.785

.687

.785

.997

.687

.997



2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

Tukey HSD

Scheffe

Tukey HSD

Scheffe

Tukey HSD

Scheffe

Tukey HSD

Scheffe





Sig.


Page 15

30


-1.0439-.9011

-1.1332-1.1299

-.6692-.8088

-1.0923-.9360

-1.1816-1.1730

-.7041-.8519

-2.2582-2.1071

-.7582-1.6123

-.0686-1.0738-2.3252-2.1554

-.8252-1.6720

-.1170-1.1335

-.3799-.2548-.1692-.1959-.1413-.2932-.3921-.2636-.1814-.2067-.1501-.3040-.5514-.4199-.3014-.3670-.1953-.3924-.5704-.4335-.3204-.3839-.2089-.4093



2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

Tukey HSD

Scheffe

Tukey HSD

Scheffe

Tukey HSD

Scheffe

Tukey HSD

Scheffe





Lower Bound

95%Confidence

Interval


Page 16

31


1.1332.6692

1.0439.8088.9011

1.12991.1816

.70411.0923

.8519

.93601.1730

.7582

.06862.25821.07382.10711.6123

.8252

.11702.32521.13352.15541.6720

.1692

.1413

.3799

.2932

.2548

.1959

.1814

.1501

.3921

.3040

.2636

.2067

.3014

.1953

.5514

.3924

.4199

.3670

.3204

.2089

.5704

.4093

.4335

.3839



2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

1-Do not expect B&W

2-DK

3-Expect B&W

Tukey HSD

Scheffe

Tukey HSD

Scheffe

Tukey HSD

Scheffe

Tukey HSD

Scheffe





Upper Bound

95%Confidence

Interval


Page 17

32

Homogeneous Subsets


16 4.562528 4.607165 4.7231

.91616 4.562528 4.607165 4.7231

.923

expect32-DK1-Do not expect B&W3-Expect B&WSig.2-DK1-Do not expect B&W3-Expect B&WSig.

Tukey HSDa,b

Scheffea,b

N 1Subset

Means for groups in homogeneous subsets are displayed.Based on Type III Sum of SquaresThe error term is Mean Square(Error) = 2.133.

Uses Harmonic Mean Sample Size = 26.409.a. Alpha = .05.b.


28 3.750016 4.500065 4.7692

.16528 3.750016 4.500065 4.7692

.192

expect31-Do not expect B&W2-DK3-Expect B&WSig.1-Do not expect B&W2-DK3-Expect B&WSig.

Tukey HSDa,b

Scheffea,b

N 1Subset

Means for groups in homogeneous subsets are displayed.Based on Type III Sum of SquaresThe error term is Mean Square(Error) = 4.095.


Page 18

33


28 .257165 .313816 .3625

.55428 .257165 .313816 .3625

.584

expect31-Do not expect B&W3-Expect B&W2-DKSig.1-Do not expect B&W3-Expect B&W2-DKSig.

Tukey HSDa,b

Scheffea,b

N 1Subset

Means for groups in homogeneous subsets are displayed.Based on Type III Sum of SquaresThe error term is Mean Square(Error) = .136.



28 .700065 .812316 .8250

.70828 .700065 .812316 .8250

.730

expect31-Do not expect B&W3-Expect B&W2-DKSig.1-Do not expect B&W3-Expect B&W2-DKSig.

Tukey HSDa,b

Scheffea,b

N 1Subset

Means for groups in homogeneous subsets are displayed.Based on Type III Sum of SquaresThe error term is Mean Square(Error) = .327.


Profile Plots


Page 19

34

Color footage Black and white footage

Condition the participant experienced.

4.00

4.20

4.40

4.60

4.80

5.00

5.20

5.40

Estim

ated

Mar

gina

l Mea

ns

expect31-Do not expect B&W2-DK3-Expect B&W

Estimated Marginal Means of To what extent did you feel mentally immersed in the film?

Page 20

35

1-Do not expect B&W

2-DK 3-Expect B&W

expect3

4.00

4.20

4.40

4.60

4.80

5.00

5.20

5.40

Estim

ated

Mar

gina

l Mea

ns

Condition the participant

experienced.Color footageBlack and white footage

Estimated Marginal Means of To what extent did you feel mentally immersed in the film?


Page 21

36



4.00

4.50

5.00

Estim

ated

Mar

gina

l Mea

ns


Estimated Marginal Means of I paid more attention to the film environment than I did to my own thoughts.

Page 22

37

1-Do not expect B&W

2-DK 3-Expect B&W

expect3

4.00

4.50

5.00

Estim

ated

Mar

gina

l Mea

ns



Estimated Marginal Means of I paid more attention to the film environment than I did to my own thoughts.


Page 23

38



0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

Estim

ated

Mar

gina

l Mea

ns


Estimated Marginal Means of COMPUTE recallf = mean(refugees,cities,camps,roosevel,godbless) (COMPUTE)

Page 24

39

1-Do not expect B&W

2-DK 3-Expect B&W

expect3

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

Estim

ated

Mar

gina

l Mea

ns



Estimated Marginal Means of COMPUTE recallf = mean(refugees,cities,camps,roosevel,godbless) (COMPUTE)

COMPUTE recallv = mean(yankdood,outfits,quaranti,usoskirt,partroop)(COMPUTE)

Page 25

40



0.40

0.60

0.80

1.00

1.20

1.40

Estim

ated

Mar

gina

l Mea

ns


Estimated Marginal Means of COMPUTE recallv = mean(yankdood,outfits,quaranti,usoskirt,partroop) (COMPUTE)

Page 26

41

1-Do not expect B&W

2-DK 3-Expect B&W

expect3

0.40

0.60

0.80

1.00

1.20

1.40

Estim

ated

Mar

gina

l Mea

ns



Estimated Marginal Means of COMPUTE recallv = mean(yankdood,outfits,quaranti,usoskirt,partroop) (COMPUTE)

Page 27

42

Multivariate Tests Table 1

Effect Value Sig. Partial Eta

Squared

Observed Power

Condition (main effect)

Pillai’s Trace Wilk’s Lambda Hotellings’ Trace Roy’s Largest Root

.147

.853

.172

.172

.003

.003

.003

.003

.147

.147

.147

.147

.919

.919

.919

.919 Expect3 (main effect)


.075

.926

.079

.067

.454

.455

.455

.155

.037

.038

.038

.063

.449

.449

.449

.506 Condition*Expect3 (interaction)


.137

.865

.155

.141

.068

.064

.060

.009

.069

.070

.072

.124

.776

.784

.791

.855

43

ANOVA TABLES Table 2 DV1 (Q2): To what extent did you feel mentally immersed in the film? Means Sum of

Squares Df Mean

Squares F Sig. Power

Condition 1=Color footage, 2=Black-and-white footage

5.042 4.256

11.848 1 11.848 5.554 .020 .646

Expect3 1=Do not expect B&W, 2=DK, 3=Expect B&W

4.532

4.643 4.772

1.107 2 .553 .259 .772 .090

Condition*Expect3 2.699 2 1.350 .633 .533 .153 Error - 219.739 103 2.133 - - - Total - 2609.00 109 - - - - Table 3 DV2 (Q24): I paid more attention to the film environment than I did to my own thoughts. Means Sum of

Squares Df Mean



4.644 4.135

4.974 1 4.974 1.214 .273 .194


3.762

4.571 4.834

21.372 2 10.686 2.609 .078 .509

Condition*Expect3 4.033 2 2.017 .492 .613 .129 Error - 421.807 103 4.095 - - - Total - 2627.00 109 - - - -

44

Table 4 DV3 (V125): Recallf = mean(refugees,cities,camps,roosevel,godbless) Means Sum of

Squares Df Mean



.452 .187

1.353 1 1.353 9.965 .002 .878


.241

.402 .316

.263 2 .131 .968 .383 .214

Condition*Expect3 1.155 2 .577 4.253 .017 .732 Error - 13.980 103 .136 - - - Total - 26.040 109 - - - - Table 5 DV4 (V126): Recallv = mean(refugees,cities,camps,roosevel,godbless) Means Sum of

Squares Df Mean



.970 .589

2.793 1 2.793 8.531 .004 .825


.660

.879 .799

.562 2 .281 .858 .427 .194

Condition*Expect3 3.301 2 1.650 5.042 .008 .807 Error - 33.717 103 .327 - - - Total - 105.280 109 - - - -

45

Condition*Expect3 Estimated Marginal Means of To what extent did you feel mentally immersed in the film? Condition*Expect3

Condition Expect3 Mean Color footage 1=Do not expect B&W 2=DK 3=Expect B&W

4.882 5.286 4.958

Black and white 1=Do not expect B&W Footage 2=DK 3=Expect B&W

4.182 4.000 4.585

Estimated Marginal Means of I paid more attention to the film environment than I did to my own thoughts. Condition*Expect3


3.706 5.143 5.083


3.818 4.000 4.585

Estimated Marginal Means of Recallf = mean(refugees,cities,camps,roosevel,godbless) Condition*Expect3


.318

.714

.325 Black and white 1=Do not expect B&W Footage 2=DK 3=Expect B&W

.164

.089

.307

Estimated Marginal Means of Recallv = mean(refugees,cities,camps,roosevel,godbless) Condition*Expect3


.847 1.314 .750


.473

.444

.849

46

A Multivariate Analysis of Variance (MANOVA) was used to assess the main and

interaction effects of the independent variables ‘condition’ (color or black-and-white) and ‘I

expected the film to be in black-and-white’ (recoded into a 3-level categorical variable named

expect3). Multivariate tests (Table 1) show a statistically significant main effect for viewing

condition and a statistically significant interaction for condition by expect3. For the main effect of

viewing condition - Pillai’s Trace p=.003 , Wilk’s Lambda p=.003 , Hotelling’s Trace p=.003 and

Roy’s Largest Root p=.003. This main effect has high statistical significance and high power. For

the interaction of condition by expect 3 – Roy’s Largest Root p=.009 (Pillai’s Trace p=.068,

Wilk’s Lambda p=.064, and Hotellings’ Trace p=.060 almost reached significance). This

interaction has near-statistical significance and high power.

Looking at the individual ANOVA tables, it can be seen which of the four dependent

variables differ across condition. There were no condition differences in I paid more attention to

the film environment that I did to my own thoughts (see Table 3). To what extent did you feel

mentally immersed in the film differed significantly by condition (p=.020) with those viewing

color footage feeling the most mentally immersed. Also, recall (both factual recall and visual

recall) differed significantly by condition (see Tables 4 and 5). Those viewing color footage had

better recall.

Since the overall MANOVA tests found a statistically significant interaction, we are

justified in looking at the individual ANOVAs to see how each dependent variable differs across

the interaction – condition by expect3. The interaction seems to be derived from the two recall

variables. Dependent variable, recallf, (F = 4.253, p=.017) and dependent variable, recallv (F =

5.042, p=.008) both show statistical significance for the interaction. The other two dependent

variables did not show an impact of the interaction. Looking at the Estimated Marginal Means and

47

the corresponding profile plots for recallf, those in the color footage condition who expect

(m=.325) and don’t expect B&W (m=.318) recall about the same amount of factual information,

those in the black-and-white viewing condition who expect B&W (m=.0307) recall a little more

factual information than those who don’t expect B&W (m=.164). The real finding here is that

those who don’t know what to expect and who are in the color footage condition (m=.714) recall a

lot more than those in the B&W footage condition (m=.089). Not having any color or B&W

expectations has a real effect on how color vs. black and white footage leads to recall of factual

information. Similarly, looking at the Estimated Marginal Means and the corresponding profile

plots for recallv, in the color footage condition, those who don’t know what to expect had the

highest recall of visual information (m=1.314) followed by those who don’t expect B&W

(m=.847) and lastly those who expected B&W (m=.750). In the black and white footage

condition, those who don’t know what to expect have the lowest recall for visual information

(m=.444), don’t expect B&W was similar (m=.473), and expect B&W had the highest recall for

visual information (m=.849). Overall, then, expectations regarding color vs. B&W makes a

difference in how viewing color vs. B&W film footage relates to recall (both types).

In conclusion, the MANOVA shows that condition does make a difference in mental

immersion and recall, and condition by expect3, especially those who don’t know whether to

expect B&W footage, makes a significant difference in recall of both factual and visual

information.

48

MANOVA - academic.csuohio.edu · MANOVA Tracy Markis COM 531 Spring 2009 The effect of color vs....

Documents

Transcript of MANOVA - academic.csuohio.edu · MANOVA Tracy Markis COM 531 Spring 2009 The effect of color vs....