Vous êtes sur la page 1sur 30

Chapter 9

Hypothesis Testing: ChiSquare tests and the one way


analysis of variance

Chi-square test for two ways


tables
Hyphothesis test for the difference in
the proportion of succeses in two or
more groups
Relationship between two categorical
variabels in a two-way crossclassification table.

Counts of particels found cross


classified by wafer condition
Wafer condition
Good

Bad

Total

Particles

Yes

14

36

50

Present

no

320

80

400

Total

334

116

450

Wafer condition
Good

Bad

Total

Particles

Yes

37.11

12.89

50

Present

no

296.89

103.11

400

Total

334

116

450

The null hyphothesis show that no


relationship between column variable
and row variabel
The alternative hypothesis show that
have difference relationship between
column variable and row variabel

The critical value of chi-square is equal to


3.481. Because 62.81 > 3.841. It reject the
null hypothesis

9.2 One way analysis of


variance ( ANOVA )
Testing for the differences
among the means of more
than two groups

Evaluating differences between


gropus as a one factor experiment ( a
completely randomized design ) in
which the variable defining the
groups is called the factor of interest
the factor of interest can have
several numerical levels and
categorical lebels

One-way ANOVA
Comparing the differences among the
population means of more than two groups
for a one factor

Sum of square total (SST) is the sum of the


squared the differences between each
individual alue and the mean of all the value
SST= Sum of( Each value- Mean of all value) 2
Sum of squares within groups (SSW) is withingroup variation that measure the difference,
mean, and sum the squares over all groups
SSW= Sum of [ (Each value in the group
Group mean)2 ]

The three variance of


ANOVA
There are three different variances : the variance

among the groups, the variance within groups and


the total variance
The variance are reffered to in yhe analysis of
variance terminology as mean square
The mean square among groups (MSA) is equal to
the sum of squares among groups (SSA) divided by
the number of groups minus 1
The mean square within group (MSW) is equal to the
sum of squares within groups (SSW) divided by the
sample size minus the number of groups
The mean square total (MST) is equal to the sum of
squares total (SST) divided by the sample size minus
1

Hypothesis testing
Z and t Tests

8.1 Testing for the difference


between two porpotions
The sample statitics need to analyze
these diffrences are te propotion of
occurrenses in group 1 and group 2
Normal distribution

Wafer condition
Good

Bad

Total

Particles

Yes

14

36

50

Present

no

320

80

400

Total

334

116

450

A level of significance of = 0.05


The proportion of good wafers without particles is 320/334 =
0.9581
The proportion of bad wafers without particles is 80/116 = 0.6897
H0 :
Ha :

1,9
6

+1,9
6

Reject H0 if Z < -1.96 or if Z > +1.96


Because Z = 7.93 is greater than the
upper critical value of +1.96, you will
jecet the null hypothesis

Testing for the difference between the


means of two independent groups
Pooled-Variance t-Test is test of
hypothesis that involved two
independent groups
Concept: the sample variances of each
group be combined (pooled) into the
one estimate of the variance common
in the two groups

Pooled variance t Test


Assumptions
Normally distribution with equal
variances
So, pooled variance t Test will not
sensitive to moderate departures
from this assumptions
The sample sizes large
To check Pooled variance t Test
Assumptions use box and whisker
plot

In normally distribution , The lower


tail is longer than the upper tail
If data didnt from normally
distribution you can use a
nonparametic procedure like
wilcoxon rank sum and separatedvariance t-test

Example

The paired t-Test


There are two approaches that inolve
releted data between grops
In first approach, you pair or match items
under study according to some other
variabels.
In second approch, repeated
measurements are obtained from the
same set of items or individuals that will
be have alike if trated alike.
In either approach about the difference
between the value of the paired item
rather rhan the values themselves

Vous aimerez peut-être aussi