Difference Between Null and Alternative Hypothesis

Ronald A. Fisher, a statistician, has been a great contributor to the field of hypothesis testing. In his book, The Design of Experiments, he has explained the usage of 'null hypothesis'. On the other hand, Jerzy Neyman and Karl Pearson, are responsible for the development of 'alternative hypothesis'.

Hypothesis is a word is commonly used in mathematics, statistics, and science. If you look up the dictionary, it states: A tentative insight into the natural world; a concept that is not yet verified but that if true would explain certain facts or phenomena.

Hypothesis is based on facts and evidence that are available, and may not necessarily be correct. It is also used in other areas, such as hypothesis of a detective about a crime, about behavior patterns, etc.

The major fields or areas where null and alternative hypothesis are used are scientific research and testing and the legal system. In the former, a prediction is made and it is gauged whether it is true or not.

In the later field, it is a generally accepted fact that a defendant is innocent until proven guilty. These and other applications of hypothesis are vital to establish facts and findings in many fields.

Among the several types of hypothesis, like inductive and deductive, simple and composite, parametric and non-parametric, we will discuss the null and alternative hypothesis with the help of a few examples that are from daily life. But first, let's try to understand the meaning of terms through their definition.

Definition: It is a hypothesis where it has been tested and seen that there exists no relationship between two variables, or it is insignificant enough to be considered as null and void. There is no alternative hypothesis, if the null hypothesis fails to get rejected.

Definition: It is a proposition which is accepted, if the null hypothesis is rejected. However, there must be enough evidences available to challenge the null hypothesis.

To understand the difference between them, let us take a look at a few examples which will give us a clear understanding of both concepts.

Consider a hypothesis test of Mr. Bob, who has to make a decision of investing either in the stocks of Company A or Company B.

Null hypothesis will establish no relationship between the two variables―stocks and returns. There is no difference in the returns given by stocks of Company A and Company B.

According to Alternative hypothesis, Mr. Bob, invests in both stocks, and he refutes the null hypothesis by his own experience. Investing in stocks of Company A will give higher returns than that of Company B

In a school, research is carried out to verify the impact on the IQ of those students who drink milk regularly.

According to Null hypothesis, there is no difference in the IQ level of children who drink milk regularly than those children who do not.

According to Alternative hypothesis, children who drink milk regularly have a higher IQ level than those who do not.

Typically used in medical research, suppose we need to test whether consumption of a certain fruit leads to any kind of allergies in the human body.

According to Null hypothesis, consumption of the fruit does not lead to any allergic reaction in the human body.

According to Alternative hypothesis, consumption of the fruit leads to an allergic reaction of red rashes on the body.

A little bit of inspiration from Sherlock Holmes. A person has been shot through the head, and the detective has to find the murderer. He suspects Mr. Ron to be the murderer and carries out a search at his house.

According to Null hypothesis, Mr. Ron is not the murderer.

According to Alternative hypothesis, Mr. Ron is probably the murderer, as a revolver was found at his house.

The search at his house, provided some evidence to refute the null hypothesis! A detective/scientist will formulate an hypothesis, and try to bring out evidences to challenge the null hypothesis.

An alternative hypothesis is, however, prone to errors―classified as Type 1 and Type 2 errors. In our day-to-day life, we encounter many such examples of hypothesis, for which we do identify the null and alternative hypothesis.

The usage of hypothesis testing in statistics, science, medicine, intelligence service, economics, etc., is widespread, and forming a null and alternative hypothesis acts as the first step of hypothesis testing.

However, as a layman, we unknowingly formulate many null and alternative thesis within our minds. If you remember your mother trying to find evidence of whether you have finished the cookies in the jar, she was just trying to form an alternative hypothesis to the null hypothesis―you did not finish the cookies!