Vous êtes sur la page 1sur 17

Analysis of Variance

(ANOVA)
Why ANOVA?
• In real life things do not typically result in two groups
being compared
• Two-sample t-tests are problematic
– Increasing the risk of a Type I error
– At .05 level of significance, with 100 comparisons, 5 will show
a difference when none exists (experimentwise error)
– So the more t-tests you run, the greater the risk of a type I
error (rejecting the null when there is no difference)
• ANOVA allows us to see if there are differences
between means with an OMNIBUS test
What is ANOVA?
• ANOVA is an abbreviation for the full name of the
method: Analysis Of Variance
• ANOVA is used to test the significance of the difference
between more than two sample means and to make
inferences about whether our samples are drawn from
population having same means.
• ANOVA is comparison of means. Each possible value of
a factor or combination of factor is a treatment.
Application of ANOVA
• Product testing, ad copy testing and concept testing are
some common applications, though ANOVA Analysis
Surveys can be used in retail environments or simulated
lab-type environments.
• We can manipulate certain variables (like promotion, ad
copy, display at the point of purchase), and observe
changes in other variables (like sales, or consumer
preferences, behavior or attitude).
ANOVA Hypotheses
• The Null hypothesis for ANOVA is that the means for all
groups are equal

• The Alternative hypothesis for ANOVA is that at least


two of the means are not equal.
• The test statistic for ANOVA is the ANOVA F-statistic
Fundamental Concepts
• You are able to compare MULTIPLE means
• Between-group variance reflects differences in
the way the groups were treated
• Within-group variance reflects individual
differences
• Null hypothesis: no difference in means
• Alternative hypothesis: difference in means
F-Ratio

Between/Within Groups
• Variance can be separated into two major
components
– Within groups – variability or differences in
particular groups (individual differences)(SSE)

– Between groups - differences depending what


group one is in or what treatment is received (SSTR)
*Steps involved calculating and interpreting F-ratio
SST=SSW +SSB

Bottom Line
• We are examining the ratio of differences
(variances) from treatment to variances from
individual differences
• If the ratio is large there is a significant impact
from treatment.
• We know if a ratio is “large enough” by
calculating the ratio of the MSB to MSW and
conducting an F test.
EXAMPLE #1
25mg 50mg 100mg
3 5 6
 
5 5 10
3 1 6
1 5 6

µ=3 µ=4 µ=8


SS=8 SS=12 SS=16
SS(within):36
Df(between):2
Total SS:92
Df(within): 9
SS(between):56
Sources SS DF S^2 Computed
f

BETWEEN 56 2 28
7
WITHIN 36 9 4

TOTAL 92
Example #2

MACHINE
1 2 3
17.6 16.2 20.1
16.9 19.6 16.1
15.8 17.8 17.8
18.4 16.1 18.6

Treatments
Levels: OBSERVATIONS: TOTALS AVERAGES:
(Machines) :
1 17.6 16.9 15.8 18.4 68.7 17.175
(309.76) (285.61) (249.64) (338.56)

2 16.2 19.6 17.8 16.1 69.7 17.425


(262.44) (384.16) (316.82) (228.01)
3 20.1 16.1 17.8 18.6 72.6 18.150
(404.01) (259.21) (316.84) (345.96
a=3,n=4 y=211 µ=17.5833

SST=22.1567
SSTR=2.0517
SSE=SST-SSTR=22.1567-2.0517
SOURCES OF SUM OF DEGREES OF MEAN COMPUTED F
VARIATION SQUARES FREEDOM SQUARE
TREATMENTS(BE 2.0517 2
TWEEN)
ERROR(WITHIN) 20.105 9

TOTAL 22.1567

5. Critical Region: We reject Ho if Fc>Fa,v1,v2 =>
   F0.05,2,9=4.26
6. Decision: Since Fc=0.459 is not greater than 
 F0.05,2,9=4.26, then we fail to reject Ho

Vous aimerez peut-être aussi