You are on page 1of 15

Hypothesis Testing

 Hypothesis is a proposed statement formulated based on


assumptions for further investigation which may or may not be
true.
 Hypothesis is an assumption with which the researcher starts
his/her investigation.
 It can be derived from personal experience, knowledge, scientific
theories, similarities or analogies in disciplines, culture and
existing literature.
 It is an assumption regarding the probable direction of outcomes.
 Hypothesis is always in declarative form.
 A hypothesis is written in such a way that it can be proven or
disproven by valid and reliable data.
 It must be formulated in simple, clear and declarative form.
 It must be clear, specific, precise, testable, relevant and must be
Criteria for related to existing body of knowledge.
designing  A good hypothesis must state relations among variables and
should have a purpose.
Hypothesis  It must be preferably unidimensional i.e. Testing one relationship
among two variable at a time.
 Step-1 : Setting up a Hypothesis : It means to establish a
hypothesis to be tested.
 Null Hypothesis is a proposed statement that shows no
Process of relationship exists between independent and dependent variables
or no difference exists in parameters of sample and population.
testing the  Null hypothesis is proposed to receive a possible rejection of the
Hypothesis assumption.
 Symbolically it is always represented as H0.
 Alternate Hypothesis is opposite of Null Hypothesis.
 It means there is relationship between variables or there is a
significant difference among the sample and population mean.
 Symbolically it is represented as H1 or Ha.
 One –tailed and two tailed tests:
 One tailed test hypothesis means the statement proposes a
comparative or unidirectional relation between two variables.
 Two tailed test hypothesis proposes a relationship exists between
two variables but without predicting any specific direction.
 Step 2 : Select an appropriate test.
 It specifies how the evaluation of data can be done systematically
following a scientific approach.
 Statistical analysts have developed numerous tests of hypothesis
which can be classified as parametric and non parametric tests.
 Parametric tests are applied when sample data is metric i.e. scale
of measurement is interval or ratio.
 Non-parametric tests are applied on non metric data i.e. when
scale of measurement is nominal or ordinal.
 Step 3: Choose the level of significance
 The significance level is the maximum value of probability (α ) of
rejecting a null hypothesis when it is true.
 The significance level is associated with overall confidence level of
the test, meaning higher the value of alpha, the greater is the
confidence in the test.
 Significance level shows how likely a pattern is there due to chance.
 The most common significance levels are .95,.99 or .90.
 The level of significance is computed by subtracting α from 1, i.e 1-α.
 Hence, the level of significance in 95% confidence level test is 5%.
 It means that the researcher is willing to take as much as 5% risk of
rejecting the null hypothesis when it is actually true.
 Therefore, the confidence with which a researcher a researcher
rejects or accepts a null hypothesis depends upon level of
significance.
 Step 4- Determination test statistics and critical region
 Determine a suitable test statistics and its distribution i.e. t,z, chi-
square, ANOVA.
 Now determine the value of test statistics which will lead to
rejection or acceptance of null hypothesis. The one that leads to
rejection of null hypothesis is called “critical region”.

 Step -5 Take the final decision
 For acceptance or rejection of null hypothesis, the test statistic
computed value is compared with the critical value table (table
value). For Z statistics

Rejection Level of
region significance
10% 5% 1%
One –tail +/- 1.28 +/- 1.64 +/- 2.33
Two-tail +/- 1.645 +/- 1.96 +/- 2.58
 In Hypothesis testing, the final decision is either to accept or reject
the null hypothesis. However, researcher may make a mistake in
this.
 Errors may occur due to error of commission or error of omission.
Errors in  Error of commission means the type of errors due to the
Hypothesis negligence of the researcher . In this wrong sample or sampling
design is selected, wrong data entry, or error in coding, tabulation
Testing and data computation is included.
 Error of omission is the type of error due to the mistakes done
unintentionally. These are basically out of control of resescher’s
perview.
 Type I error means the probability of rejecting null hypothesis
when it is actually true. In other words, it should not be rejected.
 It is an error of commission, happened due to false findings or
conclusions due to error in data collection, computation and
interpretation.
Type I and  It is denoted by α, known as significance level.
Type II Error  Usually the significance level is set to 0.05 (5%), implying that it is
acceptable to have 5% probability of incorrectly rejecting the true
null hypothesis.
 The lower α level means the lower type I error.
 Type II error means the probability of accepting the alternative
hypothesis when it is false. In other word, alternate hypothesis
should be rejected.
 It is an error of omission.
Type II Error  It is denoted by β.

Null (Ho) is true Alternate (Ha) is true

Reject Ho Type I error (α) Correct decision

Accept Ho Correct Decision Type II error (β)

You might also like