What is Hypothesis Testing?
Science primarily recognizes two generalized theories—the ‘theory of determinism’ and the ‘theory of randomness’. Theory of randomness says that whatever exists in the universe is just because of an accident and everything exists quite randomly. On the other hand, the theory of determinism states that everything in the universe is reason-based i.e. everything can be explained in definite terms.
For example, according to the theory of determinism, the earth is at a definite and perfect distance from the sun and due to this reason, life exists on planet earth; at the same time, the theory of randomness states that it’s just a chance of occurrence that the earth is at a prefect distance from the sun and the occurrence of life on planet earth is just an accident.
Symmetry Vs Non- Symmetry
All mathematical sciences believe in the theory of determinism, i.e. everything can be explained in logical terms. In other words mathematics and statistics believe that everything in the universe is symmetrical in nature, but the theory of determinism become inapplicable in case of human beings because human behavior is highly unsymmetrical in nature. One can never predict human behavior with the certainty. This non-symmetry in human nature is the basic problem for taking effective and correct decisions in business activities. For example, if someone is planning to open a restaurant in a new market then he/she can never be sure about the response of the customers because of the non-symmetry in human responses. People may accept the new restaurant or they may reject it, but the restaurant owner can’t take the risk of rejection, so before opening the restaurant he should always try to mathematically predict the response of the customers. This can be seen as moving from non-symmetry towards symmetry.
Inferential statistics
Inferential statistics or hypothesis testing is the basic criteria for drawing the mathematical predictions about a situation. It basically concentrates on concluding a particular result about a particular situation. For this purpose, researchers formulate hypothesis about a particular situation and then apply various statistical testing techniques like Z test, t test, Chi-square test etc to test the hypothesis. And on the basis of these tests, researchers conclude about accepting or rejecting the formulated hypothesis.
Example of hypothesis testing
If we consider an example of a court case, in which the CBI arrests a person with allegations of some criminal offence. Then it’s clear that CBI believes that the person is guilty and should be punished by the court. So, according to the CBI, the Null hypothesis is that “the person is guilty”; under this situation, the alternate hypothesis will be that “the person is innocent”. Now these hypothetical statements will be tested in the court through witnesses, if majority of witnesses say that yes the person is guilty, then the court will pronounce the person guilty i.e. the Null hypothesis will get accepted, otherwise it will get rejected.
Errors
In the above given example, if the person is innocent and through wrong testing, the court announces him to be guilty (i.e. null hypothesis gets accepted even if wrong) then it simply reflects Type-1 error in the testing. On the other hand, if the person is guilty and through wrong testing, the court announces him to be innocent (i.e. null hypothesis gets rejected even if correct) then it simply reflects Type-2 error in the testing.
