Test of Significance

TEST OF SIGNIFICANCE
TEST OF SIGNIFICANCE
Introduction
• Test of significance is a statistical procedure followed to test
the significant difference between statistics and the
parameter or between any two statistics. i.e. between sample
mean and population mean or between two sample means.
Hypothesis
• Any statement made about the population
Null hypothesis
• It is the hypothesis under test. It is usually denoted by Ho.
• Null hypothesis is never proved. It is either accepted or
rejected at some level of significance.
Test of significance…
Levels of significance
• Usually we will have two levels of significance.
• 5% level of significance means that we may go wrong in 5 out of 100
occasions.
• At 1% level of significance, the error in accepting null hypothesis is
reduced i.e. we may go wrong in one out of 100 occasions.
Alternate hypothesis
• Statement contrary to null hypothesis is alternate hypothesis and is
denoted by H1.
Degrees of freedom
• The number of observations which are free to move or free to vary.
DIFFERENT TESTS OF SIGNIFICANCE
Parametric test
• A distribution will be attached to the test.
• The various parametric tests are
o Normal deviate test or large sample test or Z test
o Students t test or small sample test
o Chi-square test
o Variance – Ratio test or F-test
Non- parametric test
• It will be free from distribution and it is called distribution
free method.
TYPE I ERROR AND TYPE II ERROR
In any test, we take any four type of decisions:
• Null hypothesis may be true but we reject it by our test

which is Type I error.
• Null hypothesis may be false but we accept it which is Type II
error.
• Null hypothesis is true and we accept it which is correct

decision.
• Null hypothesis is false and we reject it which is correct
decision.
STATISTICAL PROCEDURE IN ANY TEST FOR
SIGNIFICANCE
Step 1
• Formation of null hypothesis
Step 2
• Calculation of test statistics
• Test statistics = (Statistics – Parameter) / SE of difference
= (Difference in the value of two statistics)/
SE of difference
Step 3
• Depending on the value of test statistics, we take decisions
different from test to test.
LARGE SAMPLE TEST (OR) Z TEST
• Z test is carried out, when sample size is > 30.

• When n > 30,It will follow normal distribution whose equation
is
• where, m = mean; = standard deviation

TEST OF SIGNIFICANCE -‘t’ TEST
SMALL SAMPLE TEST

• When the sample size is large and variates are normally
distributed, normal test is employed to test the significance of
differences.
• With small samples and when the degrees of freedom less
than 30, the variates are not normally distributed.
• Hence we make use of the ‘t’ distribution and ‘t’ test of
significance.
• Tables have been constructed relating to ‘t’ at different
probability levels for various degrees of freedom.
SAMPLE MEAN FROM THE POPULATION
MEAN
To test the significance of the sample mean from the population

mean
Step 1
Ho: There is no significant difference between the sample mean
and population mean
Step 2
Test statistic is given by
(Contd)
where = mean of the sample of size ‘n’ with SD ‘s’ with d.

f= (n-1)
m = population mean
Step 3
Conclusion
• If | t | < table value of t for(n-1)d.f.at 5% level.’t’. is non-
significant . HO is accepted.We denote this as t = ( ) N.S.
• If | t | > table value of t’ for (n-1) d.f. at 5% level ‘t’ is
significant. Ho is rejected.We denote this as t = ( )*.
• If | t | > table value of‘t’ for (n-1) d.f. at 1 % level t is highly
significant. Ho is rejected. We denote this
as t = ( ) **
DIFFERENCE BETWEEN TWO SAMPLE MEANS
• To test the significant difference between two sample means when the sizes are less than 30
and they are dependent (Paired ‘t’ test)
• Two samples are dependent when they have some common factor linking the observations in
the two samples.
Step 1
Ho: There is no significant difference between the two sample means
Step 2
d = Difference in the observation of the two samples
s = SD of d
Step3
Conclusion: as in the previous test (with respect to d.f.)
NON PAIRED (OR) UNPAIRED 't' TEST
• To test the significant difference between the two samples,

when the samples are independent (Non-Paired or Un-paired ‘t’
test)
• By independence of the two samples, we mean that there is no
relationship between individuals contributing the sample. Thus,
the sample drawn from different populations or different parts of
the same population will be independent.
Step 1
Ho: There is no significant difference between the sample
means.
(Contd)
Step 2
Test statistics ‘t’ is given by
Step3
Conclusion
as in the previous test (with respect to d.f.)
SIGNIFICANCE OF AN OBSERVED CORRELATION
COEFFICIENT
• To test the significance of an observed correlation coefficient
Step 1
Ho: There is no significant correlation
Step 2
Step3
Conclusion: as in the previous test (with respect to d.f.)

Test of Significance

Uploaded by

Copyright:

Available Formats

Test of Significance

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Test of Significance

Uploaded by

Copyright:

Available Formats

TEST OF SIGNIFICANCE

In any test, we take any four type of decisions:

• Null hypothesis may be true but we reject it by our test

• Null hypothesis is true and we accept it which is correct

• Z test is carried out, when sample size is > 30.

• where, m = mean; = standard deviation

SMALL SAMPLE TEST

To test the significance of the sample mean from the population

where = mean of the sample of size ‘n’ with SD ‘s’ with d.

• To test the significant difference between the two samples,

You might also like