Sunteți pe pagina 1din 42

ANOVA

Analysis of Variance
From t to F
In the independent samples t test, you
learned how to use the t distribution to
test the hypothesis of no difference
between two population means.
Suppose, however, that we wish to
know about the relative effect of three
or more different treatments?
De la testul T la F
Am putea folosi testul T ca ca s comparm dou
cte dou valorile medii ale unui numr de
caracteristici.
Aceast metod in cazul cercetrii noastre nu e
adecvat din urmtoarele motive
E dificil de luat n considerare toate combinaiile posibile.
Orice statistic care are n vedere date pariale (ca n cazul
considerrii unei combinaii de dou variabile) e mai puin
consistent ca cea care folosete totalitatea informaiilor.
Unele comparaii vor furniza rezultate eronate pentru c la
nivelul eantioanelor ipoteza va fi admis din ntmplare.
From T (STUDENT)to F(FICHER)
Avem nevoie de un test cu caracter global
care va identifica diferenele semnificative
ntre caracteristicile diferitelor srategii sau
impactul factorilor de mediu asupra
procesului de implementare.
Dac rspunsul la ntrebarea noastr va fi
negativ cercetarea ulterioar nu pateu
conduce la rezultate consistente..
Un asemenea test de semnificaie aplicabil pe
un numr mare de variabile e testul F, sau
analiza de variaie, sau ANOVA.
The logic of ANOVA
Ipoteza care se testeaz prin metoda ANOVA se
refer la existena unor diferene semnificative ntre
tendinele centrale (efectul pe care aplicarea unor
srategii l are asupra eantioanelor supuse
experimentului) i ne ateptm evident ca ipoteza
nul s fie admis.
Aceast ntrebare dei se refer la medii i gsete
rspunsul prin analiza variaiei.
Printre alte motive pentru care ne concentrm atenia
asupra variaiei este acela c dorim s testm diferena
dintre medii.
DOU SURSE ALE VARIABILITII
In ANOVA, o estimaie a variabilitii ntre grupuri e comparat
cu variabilitatea n cadrul grupurilor.
Variabilitatea ntre grupuri crora li s-au aplicat tratamente diferite
(s-au experimentat diverse strategii de promovare a profesiei,
instituiei militare) conduce la diferene intre medii cu caracter
aleatoriu i datorit efectului unor factori de mediu dac e
semnificativ desigur
variabilitatea n cadrul grupurilor se datoreaz caracteristicilor .
Diferite ale elementelor din eantion care evident au participat la
acelai experiment.
ANOVA
Within-Groups Variation
Variation due to chance.
Between-Groups Variation
Variation due to chance
and treatment effect (if any existis).
Total Variation Among Scores
Variaia ntre grupuri
Exist o variaie mare ntre medii
Mari diferene ntre medii nu se datoreaz ntmprii.
E dificil de imaginat c toate grupurile sunt eantioane
aleatoare ale aceleiai populaii.
Ipoteza nul e respins indicnd un efect al
tratamentului adic eficiena cel puin unei stategii
utilizate.
Variabilitatea n cadrul grupurilor
Exist o oarecare variaie ntre mediile grupurilor.
Totui variaia n cadrul grupurilor e mai mare
pentru fiecare din grupuri.
Cu ct variaia n cadrul grupurilor e mai mare cu
att sigurana unor ipoteze referitoare la populaie
se micoreaz.
RAPORTUL F
ANOVA (F)
Within-Groups Variation
Variation due to chance.
Between-Groups Variation
Variation due to chance
and treatment effect (if any existis).
Total Variation Among Scores
Variabilitatea intre grupuri
F
Variabilitatea in cadrul grupurilor
=
DOU SURSE DE VARIAIE
1 > F
Variabilitatea intre grupuri
F
Variabilitatea in cadrul grupurilor
=
DOU SURSE DE VARIAIE
1 = F
Variabilitatea intre grupuri
F
Variabilitatea in cadrul grupurilor
=
RAPORTUL F
ANOVA (F)
Mean Squares Within
Within-Groups Variation
Variation due to chance.
Mean Squares Between
Between-Groups Variation
Variation due to chance
and treatment effect (if any existis).
Total Variation Among Scores
F =
MS
between
MS
within
mean squares within
mean squares between
RAPORTUL F
F =
MS
between
MS
within
MS
between
=
SS
between
df
between
MS
within
=
SS
within
df
within
s
2
=
(X X)
2

n 1
Sum of Squares
Degrees of Freedom
sum of squares between sum of squares within
degrees of freedom within degrees of freedom between
RAPORTUL F
F =
MS
between
MS
within
MS
between
=
SS
between
df
between
MS
within
=
SS
within
df
within
SS
total
= SS
between
+SS
within
df
total
=df
between
+ df
within
sum of squares total
degrees of freedom total
RAPORTUL F : SS intergrup
Total global
Total number of subjects.
2 2
int ergrup
T G
SS
n N
= E
2
int
( )
group grand
ergrup
SS n X X = E
RAPORTUL F : SSintragrup
Squared group total.
Numrul de
indivizi din fiecare
grup
2
2
int ergrup
T
SS X
n
= E E
2
( )
group
inragrup
SS X X = E
RAPORTUL F : SS Total
SS
total
= X
2

G
2
N

Numr total de subieci.


) ( ) ( ) (
2
group grand group grand
total
X X X X X X SS + = E =
Un exemplu: ANOVA

VIZITAREA
A.F.T
SITE A.F.T. FILM A.F.T
X
1
X
2
X
2
X
2
X
3
X
2
LICEUL MILITAR ALBA
IULIA
12 144 81 6 36
LICEUL ,,O.GOGA 10 100 7 49 7 49
LICEUL
INDUSTRIAL,,INDEPENDE
NTA
11 121 6 36 2 4
LICEUL,,GH.LAZAR


7 49 9 81 3 9
LICEUL MILITAR
CAMPULUNG
10 100 4 16 2 4

X 50 514 35 263 20 102


MEDIA=
N
X


10 7 4
ANOVA INDEX
A=514+263+102=879;
B=(50+35+20)
2
/15=735
C=(50)
2
/5+(35)
2
/5+(20)
2
/5=825

ANOVAs index va fi:
Table nr :2

DISPERSION SS Df MS F
INTERGROUP 90 2 45 10,00
INTRAGROUP 54 12 4,5
SCORE 144 14


An Example: ANOVA
Ipoteza testat.
EXISTA DIFERENE SEMNIFICATIVE
NTRE STATEGII?
Ipoteza statistic corespunzatoare.
false. is H :
:
0
3 2 1 0
A
H
H = =
Rezultat Test

2
90 (3 1) *4, 5
0, 545
144 4, 5
=

= =
+


F(2,12)=10,00,p<0,05 Exist diferene semnidficative

4. Conclusions

}nlocuind obinem
F =
5 , 4
) 5 , 4 45 ( *
15
1 3

= 1 , 0 9

M e d i i l e d i f e r s e m n i f i c a t i v . S t a t e g i a a i n f l u e n a t i m a g i n e a t i n e r i l o r d e s p r e
i n s t i t u i a i p r o f e s i a m i l i t a r


ANOVA cu msuratori repetate

Table nr: 3

S Clasa a IX-a

Clasa a X-a Clasa XI-a Clasa a XII-a
x X
2
X X
2
x X
2
x X
2
1 6 36 9 81 12 144 11 121
2 8 64 10 100 14 196 15 225
3 5 25 6 36 10 100 11 121
4 7 49 9 81 9 81 10 100
5 4 16 8 64 10 100 9 81
6 9 81 6 36 11 121 10 100
39 271 48 398 66 742 66 748
M 6,5 8,0 11,0 11,0

Calculm A , B , C v a l u e s :
A = 2 7 1 + 3 9 8 + 7 4 8 = 2 1 5 9
B=(39+48+66+66)
2
/24=1998,375
C=[(6+9+12+11)
2
+(8+10+14+15)
2
+(5+6+10+11)
2
+(7+9+9+10)
2
+(4+8+1
0+9)
2
+(9+6+11+10)
2
]/4+2039,75
D=(39
2
+48
2
+66
2
+66
2
)/6=2089,5
Calculm SS
SS
individual
=C-B=2039,5-1998,375=41,375
SS
true
(experiment)=D-B=2089,5-1998,375=91,125

SS
residual
=(A-B)-(C-B)+(D-B)=(2159-1988,375)-[(2039,75-
1988,375)+(2089,5-1998,375)]=28,125.
SS
total
= A-B = 2159-1998,375=160,625.
SS
total
= SS
individual
+ SS
experiment
+ SS
residual
Calculam gradele de libertate:
df
individual
=n-1=6-1=5.
df
experimental
=k-1=4-1=3.
df
residual
=(k-1)(n-1)=(6-1)(4-1)=15.
df
total
= N-1=24-1=23.
calculame(MS):
MS
individual
=SS
individual
/ df
individual
= 41,375/5=8,275.
MS
experimental
=SS
experiment
/ df
experimental
=91,125/3=30,375.
MS
residual
=SS
residual
/ df
residual
= 28,125/15=1,875
calculam F pentru ANOVA cu msurtori repetate:
F=MS
experimental
/MS
residual;

F=30,375/1,875=16,2.

Table nr: 4

The source
of the
dispersion
SS df MS F F,
(p<0,05)
Individual 41,375 5 8,275
Experiment 91,125 3 30,375 16,2 3,29*
Residual 28,125 15 1,875
Total 16,625 23

The following program was designed to perform ANOVA
methods.

Enter your data for a Analysis of Variance. For this to make sense you should have several
groups of data (at least 3; maximum: 26).
Number of groups:
3

Each group includes a certain number of data items. (Often all the groups have the same number of
items, but that is not required.) What is the size (i.e., the number of items) of largest group?
(maximum: 99)
Size of largest group:
5

There is no harm is over estimating the group size: blanks will be ignored. You do need to correctly
enter the number of groups.

Data Entry: ANOVA
Enter in the below set of boxes your data for each group (order makes no difference within a
group) and then click on the Calculate Now button. Empty boxes will be ignored.
Calculate Now Clear All


Data for Group A
A
01
=
12
A
02
=
10
A
03
=
11
A
04
=
7
A
05
=
10


Data for Group B
B
01
=
9
B
02
=
7
B
03
=
6
B
04
=
9
B
05
=
4


Data for Group C
C
01
=
6
C
02
=
7
C
03
=
2
C
04
=
3
C
05
=
2




ANOVA: Results
The results of a ANOVA statistical test performed at 09:56 on 12-NOV-2007
Source of Sum of d.f. Mean F
Variation Squares Squares

between 90.00 2 45.00 10.00
error 54.00 12 4.500
total 144.0 14
The probability of this result, assuming the null hypothesis, is 0.003

Group A: Number of items= 5
7.00 10.0 10.0 11.0 12.0
Mean = 10.0
95% confidence interval for Mean: 7.933 thru 12.07
Standard Deviation = 1.87
Hi = 12.0 Low = 7.00
Median = 10.0
Average Absolute Deviation from Median = 1.20

Group B: Number of items= 5
4.00 6.00 7.00 9.00 9.00
Mean = 7.00
95% confidence interval for Mean: 4.933 thru 9.067
Standard Deviation = 2.12
Hi = 9.00 Low = 4.00
Median = 7.00
Average Absolute Deviation from Median = 1.60

Group C: Number of items= 5
2.00 2.00 3.00 6.00 7.00
Mean = 4.00
95% confidence interval for Mean: 1.933 thru 6.067
Standard Deviation = 2.35
Hi = 7.00 Low = 2.00
Median = 3.00
Average Absolute Deviation from Median = 1.80

An Example: ANOVA
Calculate the test statistic.
2
X
2
X
2
X
Grand Total: 104
Imagined Retrospective Current
7 49 12 144 8 64
6 36 8 64 10 100
5 25 9 81 12 144
6 36 11 121 10 100
T:24 146 T:40 410 T:40 408
n
T
X SS
within
2
2
E E =
| | 20 400 400 144 964
4
40
4
40
4
24
408 410 146
2 2 2
= + + =
(

+ + + + =
within
SS
An Example: ANOVA
Calculate the test statistic.
2
X
2
X
2
X
Grand Total: 104
Imagined Retrospective Current
7 49 12 144 8 64
6 36 8 64 10 100
5 25 9 81 12 144
6 36 11 121 10 100
T:24 146 T:40 410 T:40 408
N
G
n
T
SS
between
2 2
E =
67 . 42 33 . 901 400 400 144
12
) 104 (
4
40
4
40
4
24
2 2 2 2
= + + = + + =
between
SS
An Example: ANOVA
61 . 9
22 . 2
34 . 21
22 . 2
9
20
34 . 21
2
67 . 42
= = =
= = =
= = =
=
within
between
within
within
within
between
between
between
within
between
MS
MS
F
df
SS
MS
df
SS
MS
MS
MS
F
An Example: ANOVA
Determine if your result is significant.
Reject H0, 9.61>4.26
Interpret your results.
There is a significant difference in the ratings of the intensity
of unrequited love depending on when (or if) the emotion
was felt.
ANOVA Summary Table
In the literature, the ANOVA results are often summarized
in a table.
Source df SS MS F
Between Groups 2 42.67 21.34 9.61
Within Groups 9 20 2.22
Total 11 62.67
After the F Test
When an F turns out to be significant, we
know, with some degree of confidence, that
there is a real difference somewhere among
our means.
But if there are more than two groups, we
dont know where that difference is.
Post hoc tests have been designed for doing
pair-wise comparisons after a significant F is
obtained.
An Example: ANOVA
A psychologist interested in artistic preference randomly assigns a
group of 15 subjects to one of three conditions in which they view a
series of unfamiliar abstract paintings. The 5 participants in the
famous condition are led to believe that these are each famous
paintings. The 5 participants in the critically acclaimed condition are
led to believe that these are paintings that are not famous but are
highly thought of by a group of professional art critics. The 5 in the
control condition are given no special information about the paintings.
Does what people are told about paintings make a difference in how
well they are liked? Use the .01 level of significance.
Famous Critically Acclaimed No Information
10 5 4
7 1 6
5 3 9
10 7 3
8 4 3
An Example: ANOVA
State the research hypothesis.
Does what people are told about paintings
make a difference in how well they are
liked?
State the statistical hypothesis.
false. is H :
:
0
3 2 1 0
A
H
H = =
An Example: ANOVA
Set decision rule.
93 . 6
12 ) 1 5 ( ) 1 5 ( ) 1 5 ( ) 1 ( ) 1 ( ) 1 (
2 1 3 1 groups of number
01 .
3 2 1
=
= + + = + + =
= = =
=
crit
within
between
F
n n n df
df
o
An Example: ANOVA
Famous Critically
Acclaimed
No
Information
10 100 5 25 4 16
7 49 1 1 6 36
5 25 3 9 9 81
10 100 7 49 3 9
8 64 4 16 3 9
T:40 338 T:20 100 T:25 151
2
X
2
X
2
X
Grand Total: 85
N
G
X SS
total
2
2
E =
33 . 107 67 . 481 589
15
) 85 (
151 100 338
2
= = + + =
total
SS
An Example: ANOVA
Famous Critically
Acclaimed
No
Information
10 100 5 25 4 16
7 49 1 1 6 36
5 25 3 9 9 81
10 100 7 49 3 9
8 64 4 16 3 9
T:40 338 T:20 100 T:25 151
2
X
2
X
2
X
Grand Total: 85
N
G
n
T
SS
between
2 2
E =
33 . 43 67 . 481 125 80 320
15
) 85 (
5
25
5
20
5
40
2 2 2 2
= + + = + + =
between
SS
An Example: ANOVA
Famous Critically
Acclaimed
No
Information
10 100 5 25 4 16
7 49 1 1 6 36
5 25 3 9 9 81
10 100 7 49 3 9
8 64 4 16 3 9
T:40 338 T:20 100 T:25 151
2
X
2
X
2
X
Grand Total: 85
within between total
SS SS SS + =
within
SS + = 33 . 43 33 . 107
64 33 . 43 33 . 107 = =
within
SS
An Example: ANOVA
06 . 4
33 . 5
67 . 21
33 . 5
12
64
67 . 21
2
33 . 43
= = =
= = =
= = =
=
within
between
within
within
within
between
between
between
within
between
MS
MS
F
df
SS
MS
df
SS
MS
MS
MS
F
An Example: ANOVA
Determine if your result is significant.
Retain H0, 4.06<6.93
Interpret your results.
People who are exposed to different kinds
of information (or no information) about a
painting do not differ in their ratings of
how much they like the painting.