Hypothesis Testing
Hypothesis Testing
Hypothesis Testing
Statistics is not only concerned with describing a set of data through observations and experiments. The heart
of statistics is actually the creation of inferences or meaningful generalizations about a given set of data or population.
This can be done by carefully observing the behaviour of every member of the population- a procedure that in actual
practice is not feasible, especially when the population is too large or inaccessible. Since you know that small sample
are related to the population, the, valid inferences can be made using only the observations from these samples. This
process of making generalizations about the characteristics of the entire population through sample statistics is called
hypothesis testing.
Generally, a hypothesis is a tentative presupposition or an inference made in order to predict the occurrence of
a phenomenon. Specifically, a statistical hypothesis is a claim about the value of a population parameter or about the
values of several population parameters. For instance, a teacher is given a large group of 230 students. Observing the
heights of the students, the teacher claims that the average height of the students is 167 cm. This claim is called a
hypothesis. In hypothesis testing, one formulates a claim about the entire population and then validates this claim by
studying cases from a few samples. The goal, then, is to identify the likelihood that a population parameter is true, using
only a sample statistic, such as the sample mean.
HYPOTHESIS TESTING
Hypothesis testing is a systematic process following a four-step procedure that includes the following:
1. State the hypothesis to be tested. The first step in hypothesis testing involves the statement of the claim that
you want to test. For instance, you want to validate the claim that the mean number of hours of sleep of college students
is 5 hours.
2. Set the standard that describes whether the claim is true or not. For example, in order to validate the claim
that the mean number of hours of sleep of college students is 5. Then majority of the chosen samples should have a
mean number of hours of sleep that is equal to or close to 5. This is the criterion.
3. Compute the test statistic. After the sample is selected, compute the test statistic, which is usually the mean.
For instance, choose 20 college students at random and compute the mean number of hours they sleep every day.
4. Make the decision. This step involves comparing the mean of the chosen sample of 20 students to be
expected mean number of hours of sleep of all college students. For example, if the sample mean is close to the
parameter, accept the validity of hypothesis. On the contrary, if a huge discrepancy is observed, the validity of the
hypothesis is refuted.
SOLUTION:
The statement ‘The mean general weighted average (GWA) of college students in a particular university is
84.8’ is an example of a null hypothesis because it describes a value that is equal to the population parameter.
The related alternative hypothesis can be stated as:
a) The mean GWA of college students in a particular university is not equal to 84.8.
b) The mean GWA of college students in a particular university is greater than 84.8.
c) The mean GWA of college students in a particular university is less than 84.8.
Directional Versus Nondirectional Test.
The appropriateness of these phrases depends on the design of the hypothesis test - either be one – tailed or
two – tailed. Also known as the nondirectional test, the two-tailed test is the standard test used in many researches and
it compares the population parameter in both directions (left and right) of the bell curve. On the other hand, the
directional test, which is also called the one-tailed test, is a test that determines the relationship between the variables
in only one direction, either the left or the right tail of the bell curve.
Example 2:
The alternative hypothesis ‘The mean GWA of college students in a particular university is not equal to 84.8.’
given in the previous example illustrates a non-directional hypothesis test.
Under a directional test, the alternative hypothesis can be stated as ‘The mean GWA of college students in a
particular university is greater than 84.8’ (right tailed) or The mean GWA of college students in a particular university
is less than 84.8’(left tailed).
One - tailed Two - tailed
1. 𝐻𝑜 : 𝜇𝑜 = 𝜇 𝐻𝑜 : 𝜇𝑜 = 𝜇
𝐻𝑎 : 𝜇𝑜 < 𝜇 ( left tailed) or 𝐻𝑎 : 𝜇𝑜 ≠ 𝜇
𝐻𝑎 : 𝜇𝑜 > 𝜇 (right tailed)
2. 𝐻𝑜 : 𝜇𝑜 ≥ 𝜇
𝐻𝑎 : 𝜇𝑜 < 𝜇 ( left tailed)
3. 𝐻𝑜 : 𝜇𝑜 ≤ 𝜇
𝐻𝑎 : 𝜇𝑜 > 𝜇 (right tailed)
*Type I and type II errors are analogous to morality. Type I is like abstaining to do something that you know is right, while type II error
is comparable to doing something that you know is wrong.
Of course, you only reject the null hypothesis when it is false and you fail to reject the null hypothesis when it
is true. Doing otherwise would certainly lead to decision error. The table below summarizes the four possible outcomes
when a decision is made in hypothesis testing.
Fail to Reject Reject
Null hypothesis is true. Correct decision Type I error
Null hypothesis is false. Type II error Correct decision
Test Concerning the Population Mean.
PROBLEM SOLVING
1. A recent survey says that Filipino children spend an average of 4 hours a day playing computer games with
the standard deviation of 30 min. a random sample of 9 children is taken from a normally distributed
population of children who spend an average of 3.5 hours playing computer games. Using 1% level of
significance, would you conclude that the statement given in the survey is correct.
2. A workout program states that there is an average weight loss of 12 pounds (lb) for those who enrol in the
program for two months. Another workout program advertises greater weight loss after two months, with
standard deviation 2.5 lb. Does the second workout program offer a higher average weight loss effect? Test the
hypothesis at 𝛼 = 0.05.
3. It is assumed that the mean IQ is 𝜇=100. In a study conducted by a researcher, a sample of n=16 people had an
average IQ of 90 with a standard deviation of 15. Is the group significantly different from the regular
population? Use 𝛼 = 0.10,.