第十章 單變量變異數分析

43
第十章 單變量變異數分析 10-1 變異數分析 10-2 單因子變異數分析的設計 10-3 變異數分析的基本假設條件 10-4 單變量變異數分析 10-5 單變量變異數分析範例 10-6 單變量變異數分析範例: One-Way ANOV A 10-7 重複量數 Repeated Measures

Upload: april

Post on 09-Jan-2016

103 views

Category:

Documents


7 download

DESCRIPTION

第十章 單變量變異數分析. 10-1 變異數分析 10-2 單因子變異數分析的設計 10-3 變異數分析的基本假設條件 10-4 單變量變異數分析 10-5 單變量變異數分析範例 10-6 單變量變異數分析範例: One-Way ANOVA 10-7 重複量數 Repeated Measures. 10-1 變異數分析. - PowerPoint PPT Presentation

TRANSCRIPT

  • 10-1 10-2 10-3 10-4 10-5 10-6 One-Way ANOVA 10-7 Repeated Measures

  • 10-1 (Analysis of Variance)ANOVA ( Analysis of Variance)MANOVA(Multivariate Analysis of Variance) (ANOVA) ()()

    Y1 = X1 + X2 + X3++Xn () ()

    MANOVA ()()()

    Y1+ Y2+..+ Yn = X1 + X2 + X3++Xn () (, )

  • 10-2 , , y1+y2+= x (y (), x 1 )2: 1. 2.

    1. , (1): HSD , Newman-Keals (2) (2): Scheffe

    2., (1), (k) (2), k

  • 10-3

    , (skewness)(kcat osis), , ()

    , , + residual

    1y, Levene >= 2y, Boxs M

  • 10-4 (ANOVA)(y)(x)(y)t()ANOVAtANOVA

    t (Test) t Test 2 t2(y)x

  • 2()2uuu

  • 2ztt2

    (n 30) ---- z ---- t

    (n< 30) , ---- z ---- t

    (n< 30) ,

  • t t (Null hypothesis)t

    t t = u1 () - u2 () / u1 u2

    t crit t I (type I) a ( 0.050.01) 12degree of freedm = (N1+N2) 2 , t crit

  • tt crit tt crit Null hypothesis (u1 = u2) u1 u2 ttcrit () Null hypothesis u1= u2

    F tF

  • 10-5 A20 ~29B30 ~39C40~49Bubble51 10

  • F,05,2,12 = 3.89FF crit, 5%Ho u 30 ~ 39 20 ~ 2940 ~ 49

  • SPSS 1. ANOVA.SAV 2. Analyze General Linear Model Univariate 3. Univariate, score 4. ,scoreDependent Variable, code 5. codeFixed Factor 6. Model 7. continue, Univariate 8. contrast () 9. continue, Univariate 10. Plots 11. continue, Univariate 12. Post Hoc, code 13. , codePost Hoc Test for, Scheffe, Tukey Duncan 14. Continue, Univerate 15. Options, 16. Continue, Univariate 17. OK,

  • Univariate Analysis of Variance

    Tests the null hypothesis that the error variance of the dependent variable is equal across groups.a Design: Intercept+code

    Levenes Test , ,

    , F = 0.43, Sig, P= 0.66 > 0.05, , ,

  • Post Hoc TestscodeMultiple Comparisons

  • Post Hoc , , TurkeyScheffe, (I) code 2 (J) code 3, (I-J), , code 3 - code 2, code 2 30 ~39code 3 40 ~49, , 30 ~39 40 ~ 49

    ANOVA:Levene 30~3940~49 30~3940~49

  • 10-6 One-Way ANOVA 27(Fit)Category 1Category 2Category 3 (Fit)

  • SPSS

  • ANOVA1. SAV AnalyzeCompare MeansOne-Way ANOVAOne-Way ANOVA fit Dependent ListCategoryFactorPost HocScheffe TukeyContinueOptionDescriptive Homogeneity of variance testContinueOK

  • (Between Groups)(Within Groups)(Total)(Between Groups)(Sum of Squares)=20.222=2(Mean Square)=10.111F =3.445 p=0.048(Within Groups) (Sum of Squares)=70.444=24(Mean Square)=2.935 (Total) (Sum of Squares)= 90.667=26fit () F (F=3.445p=.048.05)(1 2 3 )fit ()

  • Post Hoc Tests

  • Scheffe 1.

  • 2.

    *P

  • 10-7 Repeated Measures(k)Repeated Measures 15

  • score1score2score3

  • ANOVA2. SAV Analyze General Linear Model Repeated Measures Repeated MeasuresWithin-Subject Factor NamefactorNumber of Levels Add Define score1score2score3OptionsfactorCompare main effectsDescriptive statistics ContinueRepeated Measures OK

  • Measure: MEASURE_1

    score1score2score3Descriptive Statistics

    score1 =5.93=1.831score2 =6.53=1.06score3 =7.33=1.234

  • Multivariate Tests(b)

    a Exact statisticb Design: Intercept Within Subjects Design: factor

  • Mauchly's Test of Sphericity(b)Measure: MEASURE_1

    Mauchlys W 0.75Greenhouse-Geisser 0.75Huynh-Feldt 0.75Mauchly .8222.55df=2p=.279>.05

  • Tests of Within-Subjects EffectsMeasure: MEASURE_1

    (Sphericity Assumed)typeIII SS=14.8df=2MS=7.4F=4.723p=.017

  • Tests of Between-Subjects EffectsMeasure: MEASURE_1Transformed Variable: Average

    (Tests of Between-Subjects Effects ) (Block)=40.133=14=2.867

  • Estimated Marginal MeansfactorEstimatesMeasure: MEASURE_1

    95%

  • Pairwise ComparisonsMeasure: MEASURE_1

    Based on estimated marginal means* The mean difference is significant at the .05 level.a Adjustment for multiple comparisons: Least Significant Difference(equivalent to no adjustments).()(Mean Difference).05 (M=7.333) (M=5.933)

  • **P