0% found this document useful (0 votes)
23 views31 pages

4.2 Different Tests

The document outlines various statistical tests including Z-statistic, t-statistic, chi-square, and F-statistic, detailing their applications for hypothesis testing in different scenarios. It provides examples of decision-making processes for accepting or rejecting null hypotheses based on test statistics and significance levels. Additionally, it discusses assumptions for t-tests, variance testing, and presents problems to illustrate the application of these statistical methods.

Uploaded by

Abu Sayeed
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
23 views31 pages

4.2 Different Tests

The document outlines various statistical tests including Z-statistic, t-statistic, chi-square, and F-statistic, detailing their applications for hypothesis testing in different scenarios. It provides examples of decision-making processes for accepting or rejecting null hypotheses based on test statistics and significance levels. Additionally, it discusses assumptions for t-tests, variance testing, and presents problems to illustrate the application of these statistical methods.

Uploaded by

Abu Sayeed
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 31

Different Tests

Application of Different tests


• Application of Z-statistic:
– Test of a single population mean
– Test of equality of two population means
– Test up to single population proportion
– Test for different between two population
proportions
– test of a specified correlation coefficient
– test of equality of two population correlation
coefficient
Application of Different tests
• Applications of t-statistic (small sample test):
– Test of a single population mean
– test of difference between two population
means
– test of significance of a correlation coefficient
with zero value
– test of a population of regression coefficient
with zero or a specified value
– test of difference between two population
regression coefficient
Application of Different tests
• Application of -statistic:
– Test of significance for population variance
– test of independence of two attributes
– test of goodness of fit
– test of equality of several variances
– test of equality of several correlation coefficient
– test of equality of several population
proportions
Application of Different tests
• Application of F-statistic
– Test of significance of difference between two
population variances
– test of significance of several population means
– test of significance of two or more regression
coefficient
t-distribution
• When we use this distribution?
– For expensive populations such as satellites,
aero-planes, nuclear reactors, super computer,
etc.
• Why?
– Because the investigation of large samples in
such cases is uneconomical, impracticable and
time consuming.
T-distribution
If is a random sample drawn form a normal population with unknown
mean and variance . Also, let

Be the sample mean and unbiased estimate of population variance , then the
statistic

Is said to have t-distribution with degrees of freedom , i.e.,


symbolically,

. .

Remarks:
 The t distribution depend on 𝑛, the size of the sample
 The sample is drawn from a normal population.
 As the sample size n increases, t-distribution also tends to standard normal distribution.
∑ ̅
 𝑆 is different than the sample variance 𝑠 = . The relation between 𝑆 and 𝑠 is :
𝑛 𝑛−1
𝑆 = 𝑠 𝑜𝑟 𝑠 = 𝑆
𝑛−1 𝑛
Decision
• Two tail test:
 : we accept
 : we reject
• Right-tailed test:
 : Accept
 : Reject
• Left-tailed test:
 : Accept
 : Reject
Assumptions in t-test
• The sample is drawn from a normal
population.
• Sample observations are independent.
• Sample size is small.
• Population variance is unknown.
Problem
• An ambulance service company claims that on an average it takes
20 minutes between a call for an ambulance and the patient’s
arrival at the hospital. If in 6 calls the time taken (between a call
and arrival at hospital) are 27,18,26,15,20,32. Can the company’s
claim be accepted?
– Here,
– Let X be the time taken between a call and a patient’s arrival at
hospital. From given data


– Standard deviation,
( )
– Thus,
Solution (ctd.)
1. Null hypothesis: minutes
2. Alternative hypothesis: minutes
3. Level of significance:
4. Critical region:
5. Test statistic:
6. Decision: Accept , since
with 5 degrees of freedom.
That is, accept the claim of the company.
Exercise
1. The mean lifetime (mlt) of computers manufactured by a company is 1120 hours with
standard deviation of 125 hours.
a. Test the hypothesis that mean lifetime of computers has not changed if a sample of 8 computers has
a mlt of 1070 hours
b. Is there decrease in mlt?
Use (𝑖) 0.05 and (𝑖𝑖) 0.01 L.O.S.(Given, 𝑡 . with 7 dof ±3.499, 𝑡 . with 7 dof is ±2.365)
2. If 5 pieces of certain ribbon selected at random have mean breaking strength of 169.5
pounds with s.d. of 5.7, do they confirm to the specification mean breaking strength of 180
pounds?
3. In a random sample of 10 bolts produced by a machine the mean of bolt is 0.53 mm and
standard deviation 0.03 mm. can we claim from this that the machine is in proper working
order if in the past it produced bolts of length 0.55 mm? Use (a) 0.05 and (b) 0.01 L.O.S.
4. An auditor claims that he takes on an average 10.5 days to file income tax returns (I.T.
returns). Can this claim be accepted if a random sample shows that he took
13,19,15,10,12,11,14,18 days to file I.T. returns? Use (a) 0.01 (b)0.05 L.O.S.
5. An automobile tyre manufacturing claims that the average life of a particular grade of tyre is
more than 20,000 kms when used under normal driving conditions. A random sample of 16
tyres was tested and mean and S.D. of 22000 and 5000 kms respectively were computed.
Assuming the life of the tyre in kms to be normally distributed, decide whether the
manufacturer’s claim is true? Use 5% level of significance.
distribution
• If are independent normal variates with mean
and variance then

are independent normal variates with mean 0 and variance


1. Then the sum of squares of the variates i.e.,

Is distributed as with n degrees of freedom. Symbolically,


distribution
• If are n independent normal
variates with mean zero and standard
deviation unity then is a
random variate having distribution with
probability density function given by

where
and constant depending on the
degrees of freedom.
Properties
• The distribution of lies in the first quadrant.
• The range of distribution is form 0 to .
• It has a unimodal curve.
• distribution has only one parameter n which is its degrees of
freedom
• The shape of the curve greatly vary as n varies.
• The mean and variance of the distribution with n are:

• If and are two independent variates with and degrees of


freedom respectively, then their sum will be distributed as chi-
square with This addition property of chi-square holds
good for any number of variates. Thus, if are
independent variates with Respectively, then
their sum with
Test of significance for
population variance
• Let a random sample of size be drawn from a
normal population with mean and variance ,
and being unknown. The test statistic is

in terms of or , the statistic in the above


equation can be written as,

Where,
Decision
• In case of two tail alternative, reject at level of significance if
, ,

• In case of a right-tailed alternative, reject at level of significance if


,
• In case of a right-tailed alternative, reject at level of significance if
,
Distribution curve
Problems (Variance)
• The variability in the tensile strength of a type
of steel wire must be controlled carefully. A
sample of the wire is subjected to test and it is
found that sample variance is . The
sample size was observations.
a. Test the hypothesis that the population variance is
25 against the alternative that the variance is
greater than 25. Use a 0.05 level of significance.
b. Construct the 95 percent confidence limits on the
variance in the tensile strength of the wire.
Problems(Variance)
• Solution: We have to decide between the hypotheses
𝐻 : 𝜎 = 25 𝑎𝑛𝑑 𝐻 : 𝜎 > 25
• The level of significance is 𝛼 = 0.05

• The test statistic is 𝜒 = which under 𝐻 , has a 𝜒 − distribution with (𝑛 − 1)


degrees of freedom, assuming that the population is normal.
• The critical region is 𝜒 > 𝜒 . , = 25.0 (one-tailed test)
• We calculate the value of 𝜒 from the sample data as
𝑛𝑠 16(31.5)
𝜒 = = = 20.16
𝜎 25
• Conclusion: since the calculated value of 𝜒 falls in the acceptance region, so we accept
our null hypothesis, i.e., we have reasonable evidence to conclude that 𝜎 = 25
2nd Part:

• The 95 percent confidence interval for 𝜎 = 25 is given by <𝜎 <


. , . ,

Substituting the values, we get


16 31.5 16 31.5
<𝜎 <
27.49 6.26
𝑜𝑟, 18.3 < 𝜎 < 80.5
Hence the desired confidence limits on the population variance are (18.3, 80.5)
Problems(Variance)
• The following are the weights (in gram) of a randomly selected
sample of 11 apples in a shop.

The weight of apples follows normal distribution with


mean and variance . Can we conclude that the population
variance of apples of the shop is more than 50
• The daily duration of telephone calls received by the enquiry
department of a small industry for a randomly selected 11 days
over a quarter are as follows:
160,172,121,144,100,108,175,200,105,95,102
The manager of industry says that the population variance of the
daily duration of calls over the quarter is 1500. The authority thinks
that it is overestimated. How would you comment on the variance?
Definitions
• Contingency table: Data for attributes arranged in
two-directional tabular form for the test of
independence is called contingency table. The order
of the table is determined by the number of
categories of two attributes.
• Observed Frequency: The frequencies obtained by
observation. These are the sample frequencies.
• Expected Frequency: The frequency corresponding to
a particular cell obtained by dividing the product of
row total and column total passing through that cell
by the total frequencies.
Test of hypothesis about independence of two
attributes( - Non parametric test)
• Test statistic:

Where, degrees of freedom


Problem
• A certain drug is claimed to be effective in curing cold. In
an experiment on 500 persons suffering from cold, half of
them were given the drug and half of them were given
the sugar pills. The reaction to the treatment on patients
are recorded as in the following table:
Helped Harmed No effect Total
Drug 150 30 70 250
Sugar pills 130 40 80 250
Total 280 70 150 500

On the basis of the information can it be concluded that


there is a significant difference in the effect of the drug and
sugar pills.
F-distribution
Let there be two normal population with means and and variances and
respectively. The two populations can be symbolically written as and
. Let be a sample of size from the population
and be another sample of size from the population .
Then according to definition of , we can form the following two - statistics:

∑ ̅ ∑
Where, are unbiased estimators of
respectively. Now using values, we define the statistic F as

The sampling distribution of the statistic in the above expression is F-distribution


with degrees of freedom.
Tests based on F-
distribution

Test of equality of Test of equality of


two population several
variances population means
Test of equality of two
population variances
• The basic difference between t and F
distribution is in t-distribution we assume
that the population variances must be equal
while in F-distribution is used to ascertain
the equality of population variance.

• Test statistic: (among put


the bigger one on the numerator, in this case
)
Decision
Problem
• Two samples are drawn from two normal
populations. From the data information, test
whether the two samples have the same variance at
5% level of significance.
Sample 1 60 65 71 74 76 82 85 87
Sample 2 61 66 67 85 78 63 85 86 88 91

• Solution: Here, null hypothesis:


• Alternative hypothesis:
• To test , the test statistic is,
Solution
Exercise
• Given the following information about two samples
drawn from two normal populations :

Test the equality of two population variance.


• The following figures related to the number of limits
produced per shift be two workers A and B for a number
of days:
A 19 22 24 27 24 18 20 19 25
B 26 37 40 35 30 30 40 26 30 35 45

Can it be inferred that A is more a stable worker compared


to B? Answer using F-test at 5% level of significance.

You might also like