What Is a Null Hypothesis? (Plus Types, Tips and Example)

By Indeed Editorial Team

Published 26 April 2022

The Indeed Editorial Team comprises a diverse and talented team of writers, researchers and subject matter experts equipped with Indeed's data and insights to deliver useful tips to help guide your career journey.

Statistics can help you with procedures for data gathering, analysis and interpreting and proposing your findings. A null hypothesis is an important component of statistics and research in various careers, such as market research and financial analysis. Learning about null hypotheses and how they work can help you enhance your analytical, critical thinking and research skills. In this article, we explain what null hypotheses are, discuss how they work and explore an example of one.

What is a null hypothesis?

Before learning what a null hypothesis is, it's important to understand first what a hypothesis is. A hypothesis is a theory or speculation based on insufficient evidence and requires experimentation and testing. With further testing, you can prove that a hypothesis is true or false. For example, say you speculate or hypothesise that the number of working hours in a day doesn't affect employee morale. You then observed two employees: one has an eight-hour shift, and the other has a 10-hour shift. You observed them for one week and proved that your hypothesis is true.

Null hypotheses are hypotheses that say there's no statistical significance between two variables in the hypothesis. They're the hypotheses that the researcher is trying to disprove. In the example, your hypothesis might be something like this: There's no statistically significant relationship between the number of working hours in a day and the morale of employees. Null hypotheses challenge a researcher to disprove them to show that there's a significant relationship between the two variables in the hypothesis.

Related: 16 Must-Have Data Analyst Skills (Plus Job Responsibilities)

Types of null hypotheses

There are two types of null hypotheses. These include the following:


A non-directional hypothesis predicts that the cause or independent variable has an effect on the dependent variable. It doesn't specify, though, the direction of the effect. It just states that there's a difference. For example, in an experimental study, you can say, There's a difference between the number of cold symptoms experienced in the following day after exposure to a virus for employees who have been sleep-deprived for 24 hours compared to those who haven't been sleep-deprived for 24 hours.


A directional hypothesis predicts the direction of the difference among samples. The basis for this directional guess can be your knowledge base, your own experience or evidence in the professional literature. In an experimental study, for example, you may say, Employees who have been sleep-deprived for 24 hours have more cold symptoms the following day after exposure to a virus compared to employees who haven't been sleep-deprived. The directional hypothesis compares two conditions or groups and states which one has more or less or is slower or quicker.

What are the differences between null hypotheses and alternative hypotheses?

The null hypotheses state that there's no actual difference among a set of figures, conversely, the alternative hypotheses suggest there are no differences between the figures. Thus, the null hypotheses contradict the alternative hypotheses.

Analysts and statisticians develop an alternative hypothesis to explain the differences in statistical relationships or describe a set of circumstances. For example, you can say, Employees are more productive if their employers give them one break every three hours, as opposed to one break every four hours. Using the alternative hypothesis as a guide, researchers conduct studies and perform experiments to disprove and reject null hypotheses.

Related: A Guide to Analytical Skills: Definition, Examples and Tips

How do null hypotheses work?

Null hypotheses propose there are no differences between a set of variables or relationships. Also, their alternative hypotheses propose that differences among those relationships exist. Thus, researchers presume that the null hypotheses are accurate until there are sufficient and statistically significant data that prove otherwise. Researchers often use the following steps for hypothesis testing:

  1. State the two hypotheses: The first step researchers take when testing hypotheses is to state the null and alternative hypotheses so that only one can be right.

  2. Create an analysis plan: The researchers formulate an analysis plan that outlines how they evaluate the data

  3. Implement the plan: In this step, the researchers implement the plan and physically analyse the sample data.

  4. Analyse the results: In this step, the researchers analyse the results. They then reject the null hypotheses or claim that the differences they've observed are explainable by chance alone.

When testing hypotheses, researchers often use a p-value or probability value as evidence against null hypotheses. The probability value tells you how likely it is that your data may have occurred under the null hypotheses. Smaller p-values show strong statistical data and research that disproves the null hypotheses. Researchers conduct significance tests to show confidence in the null hypotheses. They also use significance testing to determine whether the data is due to chance. During testing, researchers face two scenarios:

  • Fail to reject the hypothesis: If the p-value is greater than the significance level, the results aren't statistically significant. Here, statisticians may fail to reject the hypothesis because of insufficient data, errors in the data or other parameters.

  • Reject the hypothesis: If the p-value is less than or equals the significance level, then the results support the alternative hypotheses, meaning the data is statistically significant, and researchers can reject the null hypotheses. Rejection of the hypothesis doesn't mean the experiment didn't find the required answers, but indicates a need for further experimentation to see if there's a relationship between the variables.

Related: Research Skills: Definition, Benefits and How to Develop

Tips for stating null hypotheses

Here are a few tips for stating null hypotheses:

Identify possible circumstances

When stating the null hypotheses, it's important to identify all outcomes. For example, after examining a problem and identifying questions to ask, you can conclude that the null hypotheses are the expected outcomes. Next, you can develop alternative hypotheses that work to reject the expected outcomes. In this way, you try to predict all circumstances and either reject the null hypotheses and accept the alternative hypotheses or fail to reject the null hypotheses.

Related: 16 Jobs for Analytical Thinkers (With Salaries and Duties)

Create the null hypotheses

To create the null hypotheses, consider examining the problem you're trying to solve and determine the questions you're trying to ask. Typically, the null hypotheses are direct representations of the expected outcomes. You start by asking a question, then rephrasing that question as a statement that assumes no relationship between the two variables.

Think of the hypothesis as a fact

Think of the null hypotheses as facts and the alternative hypotheses as opinions or beliefs. To state the null hypotheses, it's important to regard them as the status quo or the way things currently exist. Thus, if you accept the null hypotheses as facts, then the alternative hypotheses are the statements that dispute that fact. Researchers and statisticians aim to disprove the null hypotheses while proving the alternative hypotheses to be true.

Related: What Is Equity Research? (Plus How To Establish a Career in the Field)

Null hypothesis example

Here's an example of how you can use the hypothesis:

State the hypothesis

A school district's superintendent claims their district's high school math students receive average scores of eight out of 10 on their math tests. Here's an example of their null and alternative hypotheses:

  • Alternative hypothesis: The school district's high school math students receive average test scores that are not equal to eight out of 10.

  • *Null hypothesis: The school district's high school math students receive average test scores of eight out of 10.*

Test the hypothesis

To test the validity of the hypothesis, consider the following:

  1. Gather data: Collect data from a sample of 500 high school students from the district.

  2. *Calculate the average: Calculate the average math test score from those 500 samples.*

  3. *Compare the result: Compare the results to the original statement that the district's high school math students receive average scores of eight out of 10 on tests.*

  4. *Determine whether to reject or accept the hypothesis: Use the data to either reject the null hypotheses in favour of alternative hypotheses.*

Calculate and interpret the results

When calculating the average test scores, it's important to note that hypothesis testing assumes the conjecture is true, unless proven otherwise. Therefore, the claim that the district's high school math students receive average scores of eight out of 10 on tests provides data that suggests if eight out of 10 is the average, then the scale of outcomes can be any value from 7.2 to 8.8, as the population mean is 8.0. If the calculated average is any value outside this range, you can reject the hypothesis because the average score isn't eight out of 10.

Explore more articles