0% found this document useful (0 votes)
171 views36 pages

W8 ANOVA and Kruskall Wallis

Based on the document: 1. There were 3 groups being compared in the study. 2. The F-ratio value was 6.72. 3. The null hypothesis (Ho) was rejected, indicating that not all groups had the same median class size.

Uploaded by

Yurie Riri
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
171 views36 pages

W8 ANOVA and Kruskall Wallis

Based on the document: 1. There were 3 groups being compared in the study. 2. The F-ratio value was 6.72. 3. The null hypothesis (Ho) was rejected, indicating that not all groups had the same median class size.

Uploaded by

Yurie Riri
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 36

ANOVA & Kruskal-Wallis

ANOVA (Analysis of Variance)

Analisis varians terhadap hasil sesuatu


kajian; khususnya ke atas suatu varians
statistik bagi tujuan menentukan pengaruh
pembolehubah-pembolehubah kajian ke
atas varians yang dimaksudkan.
LATAR BELAKANG DATA & ANOVA
 Semua set data akan mempamer serakan
(variability)
 Serakan boleh wujud secara :
 Dalam kumpulan (within group)
 Antara kumpulan (between group)
 Kaedah ANOVA boleh menentukan sama ada tahap
perbezaan pada serakan tersebut signifikan atau
sebaliknya.
  Perbezaan

di antara serakan dalam kumpulan
dengan serakan di antara kumpulan boleh
memberi ketentuan mengenai :

H0 : Serakan dalam kumpulan = serakan antara kumpulan

Ha : Serakan dalam kumpulan serakan antara kumpulan


ANOVA Sehala
ANOVA Sehala
(one-way ANOVA)

Satu faktor (pembolehubah yang


dimanipulasi)
Satu pembolehubah bersandar
Tiga atau lebih kumpulan
perbandingan
One-way ANOVA ASSUMPTIONS
 Assumption of Normality
Like the t-test you can place fast and loose with this
one, especially with large enough sample size – see:
the Central Limit Theorem
 Assumption
of Homogeneity of Variance
(Homoscedasticity)
Like the t-test, this isn’t problematic unless one
level’s variance is much larger than one the others
(~4 times as large) – the one-way ANOVA is robust to
small violations of this assumption, so long as group
size is roughly equal
 Independence of Observations
Like the t-test, the ANOVA is very sensitive to
violations of this assumption – if violated it is more
appropriate to use a Repeated-Measures ANOVA
One-Way ANOVA
 Assumptions of ANOVA:
 Independence of Observations
 Homoscedasticity

 Normality
 Equal sample sizes – not technically an assumption, but effects
the other 3

 How do we know if we violate one (or more) of


these? What do we do?
One-Way ANOVA
 Independence of Observations
 Identified methodologically
 Otherthan using repeated-measures tests
(covered later), nothing you can do

 Equal Sample Sizes


 Add more S’s to the smaller group
 DON’T delete S’s from the larger one
One-Way ANOVA
 Normality
 Can identify with histograms of DV’s (IV’s are
supposed to be non-normal)
 More appropriate to use skewness and kurtosis
statistics
 Ifdetected (and sample size very small), use
appropriate transformation
One-Way ANOVA
 Estimates of Effect Size in ANOVA:
1. η2 (eta squared) = SSbetween/SStotal
 Unfortunately, this is what most statistical computer
packages give you, because it is simple to calculate, but
seriously overestimates the size of effect

SS groups   k  1 MS error
2. ω2 (omega squared) = SS total  MS error

 Less biased than η2, but still not ideal


One-Way ANOVA

 Estimates of Effect Size in ANOVA:


3. Cohen’s d = X1  X 2 2 F 2 t
 
sp df error n1  n 2  2

Remember: for d,
 d = 0.2 = small effect,
 d = 0.5 = medium, and
 d = 0.8 = large
APLIKASI ANOVA Sehala
 Boleh digunakan apabila min pembolehubah bersandar berbeza-beza di
antara kumpulan

 Contoh 1:
 Contoh 2:
 Kajian membandingkan 3 orang pensyarah yang mengajar satu bahagian
dalam satu kursus. Mereka memberi peperiksaan akhir yang sama dan mereka
ingin tahu adakah terdapat perbezaan antara skor pada bahagian-bahagian
mereka.
Apakah Analisis yang Dilakukan Menerusi
Anova Sehala ?
Pensyarah N Mean
Membandingkan 1 parameter (min skor)
Dr. A 60 68.367
di antara semua kumpulan
Dr. B 87 71.448
Dr. C 98 67.939

• Menentukan sama ada perbezaan min di antara kumpulan; signifikan,


atau tidak signifikan menerusi satu perbandingan umum (main
comparison)

• ANOVA tidak memerihalkan kumpulan khusus yang menghasilkan


perbezaan signifikan
[Bagi tujuan ini, perlu analisis post hoc]
CONTOH JADUAL ANOVA Sehala
JADUAL ANOVA Sehala
Source of Degree of Freedom Mean of Squares (MS) F-ratio
Variation
SSbetween df =

SSwithin df =
SSwithin
SStotal

SStotal

 k = # kumpulan (treatment group) xi = min untuk kumpulan i


n = # pemerhatian dalam setiap kumpulan SS = Sum of Squares
satu pemerhatian dalam kumpulan i MS = Mean of Squares
= min semua kumpulan (grand mean)
LATIHAN

1. Ada berapa kumpulan ?


2. Ada berapa nilai pemerhatian ?
3. Apakah nilai SSb?
4. Apakah nilai F?
5. Dapatkan effect size (eta quared).
6. Nyatakan keputusan Ho.
LATIHAN

1. Ada berapa kumpulan ? (k = 3) ; df = k – 1 = 2


2. Ada berapa nilai pemerhatian ? (n = 60) ; df = n – k = 57
3. Apakah nilai SSb ? (SSb = 35.73) ; SST = SSb + SSw
4. Apakah nilai F ? (F = 4.066) ;
5. Dapatkan effect size (eta quared). 0.125
6. Nyatakan keputusan Ho. – Reject Ho η2 (eta squared) = SSbetween/SStotal
Ujian Bukan Parametrik
KRUSKAL-WALLIS
Ujian Kruskal-Wallis :
Analysis of Variance by Ranks
 Digunakan untuk membandingkan 3 atau lebih sampel
untuk memastikan sampel-sampel tersebut datang
daripada populasi yang sama
 Menggunakan skala ukuran ORDINAL
 Merupakan alternatif kepada ujian ANOVA sehala
 Setiap sampel mempunyai sekurang-kurangnya 5
pemerhatian
 Sampel data disusun dari rendah ke tinggi sebagai satu
kumpulan (Rank)
Kruskal-Wallis Assumptions
 Assumption #1 : The dependent variable is measured at
continuous or ordinal level and the independent variable
consists of two or more categorical, independent groups
 Assumption #2 : You should have independence of
observations, which means that there is no relationship
between observations in each group of the independent
variable or between the groups themselves
 Assumption #3 : The distribution of scores for each group
of the independent variable have the same shape (same
variability)
Ujian Kruskal-Wallis
  Hipotesis bagi ujian Kruskal-Wallis :
 H0 : Kesemua populasi k mempunyai taburan yang sama
 Ha : Tidak semua populasi k mempunyai taburan yang sama
 Ujian statistik Kruskal-Wallis :

 Jika maka H dianggarkan mempunyai taburan (Chi kuasa dua)


Contoh
 Adakah bahagian yang berbeza mempunyai saiz kelas yang berbeza?
Class size Class size Class size
H0 : MedianM = MedianE = MedianH (Math) (English) (History)
: Ketiga-tiga bahagian mempunyai saiz kelas 23 55 30
median yang sama 41 60 40
54 72 18
Ha : Tidak semua bahagian mempunyai saiz kelas
median yang sama 78 45 34
66 70 44
Group Class size Rank
M 23
M 41
 1. Hasil tambah rank bagi kumpulan M = ?
M 54
M 78 2. Hasil tambah rank bagi kumpulan E = ?
M 66
3. Hasil tambah rank bagi kumpulan H = ?
E 55
E 60
E 72
E 45
E 70
H 30  
H 40
H 18
H 34
H 44
Group Class size Rank  1. Hasil tambah rank bagi kumpulan M =
M 23 2
2. Hasil tambah rank bagi kumpulan E =
M 41 6
M 54 9 3. Hasil tambah rank bagi kumpulan H =
M 78 15
M 66 12
E 55 10
E 60 11
E 72 14
E 45 8
E 70 13
H 30 3  
H 40 5
H 18 1  
H 34 4
H 44 7

   kira = 6.72
H

 Hkritikal ?

 Cari di Jadual A.4, taburan Chi kuasa dua

 Hkritikal = 5.992

 Oleh sebab
 H0 ditolak. Tidak semua bahagian
mempunyai saiz kelas median yang sama.

H0 ditolak jika Hkira > Hkritikal



  Further analysis (Pairwise Comparisons of Average Ranks)
 The pairwise comparison test statistic :

where and is the mean of the ranks of the observations


from population i and j

 The critical point for the paired comparisons :

Reject if
  Pairwise Comparisons of Average Ranks
 

Reject if D > CKW

|𝑅´ 𝑖 − 𝑅´ 𝑗|
 𝐷=
PENGIRAAN MELALUI SPSS

Langkah 1 : Input data


Langkah 2 : Pilih Analyze > Nonparametric Tests > Legacy Dialogs
> K Independent Samples…
Langkah 3 :
Letakkan Classsize di Test
Variable List

Letakkan Group di Grouping


Variable

Klik Define Range…

Langkah 4:
Input 1 di ruangan Minimum dan 3 di
ruangan Maximum

Klik Continue

Klik OK
 Langkah 6 : Keputusan

H0 ditolak
KUIZ

Dapatkan nilai SSw, nilai F, bil. kumpulan dan keputusan bagi Ho.

You might also like