What Is Statistical Hypothesis Testing?

Statistical hypothesis testing is a method for formulating a hypothesis about a population and probabilistically evaluating whether that hypothesis is consistent with sample data.

For example, when considering questions such as “Is this new teaching material more effective than the conventional material?”, “Does this drug really work?”, or “Is there a difference between the means?”, simply comparing sample means is not sufficient because samples contain random variation.

In hypothesis testing, If there were no difference, how unlikely would data like the present data be? Statistical decisions are made by considering this question.

What Is the Null Hypothesis?

Among the hypotheses formulated at the outset of a test, the one that researchers generally the hypothesis that one initially seeks to reject what is often set up first is null hypothesis .

The null hypothesis represents a baseline condition such as “there is no difference,” “there is no effect,” or “the means are equal.” More concretely, it can be a hypothesis that the data from the sampled group are consistent with a population mean or known reference value.

For example, if we want to examine whether a mean is 50, H?: μ = 50 For example, the null hypothesis may be stated as follows.

What Is the Alternative Hypothesis?

The hypothesis positioned as the opposite of the null hypothesis is the alternative hypothesis .

This often corresponds to what the researcher ultimately wishes to claim and is expressed in forms such as “there is a difference,” “there is an effect,” or “the means are not equal.”

Using the preceding example, H?: μ ≠ 50 It can be written in this form. Alternatively, a one-sided test may be used, such as “the mean is greater than 50” or “the mean is less than 50.”

In hypothesis testing, When the null hypothesis can be rejected, the alternative hypothesis is supported. judgment proceeds through this sequence.

“Failing to Reject the Null Hypothesis” Is Different from “Accepting” It

A particularly important point when learning hypothesis testing is that “failing to reject the null hypothesis” and “accepting the null hypothesis” are not the same thing This is the key point.

Hypothesis testing assumes that the null hypothesis is true and examines how unlikely the observed data would be under that assumption. Therefore, inability to reject the null hypothesis does not mean that the null hypothesis has been proven true.

Simply It means only that the present data did not provide sufficient evidence to reject the null hypothesis. This distinction is very important.

Basic Step 1 of Hypothesis Testing: Formulate the Hypotheses

In the first stage of hypothesis testing, Formulate both a null hypothesis and an alternative hypothesis. 。

The content the researcher wants to argue is generally placed in the alternative hypothesis. For example, if the claim is that “the new method improves scores,” then “there is no difference” is set as the null hypothesis and “the new method is better” as the alternative hypothesis.

What matters at this stage is It is important to clarify what is being compared and in which direction a difference is expected. .

Basic Step 2 of Hypothesis Testing: Calculate the p-Value

Next, assuming that the null hypothesis is true, calculate the probability of obtaining the observed data, or data even more extreme than the observed data is calculated. This is the p-value .

The smaller the p-value, the more the data are considered unusual if the null hypothesis is true. Conversely, a large p-value means that the observed data are not particularly unusual under the null hypothesis.

The p-value is therefore an extremely important basis for judgment in hypothesis testing.

Basic Step 3 of Hypothesis Testing: Compare with the Significance Level

Finally, compare the p-value obtained in Step 2 with significance level the specified threshold.

The significance level is the threshold below which a result is regarded as sufficiently unusual. A commonly used level is 5% and 1% is commonly used.

If the p-value is less than or equal to the significance level If the p-value is at or below this level, the observed data are considered unlikely under the null hypothesis, and the null hypothesis is reject the null hypothesis rejected. The alternative hypothesis is then supported.

If the p-value is greater than the significance level, the null hypothesis cannot be rejected. In that case, the hypothesis test has not produced a statistically significant result.

To summarize the hypothesis-testing process,

The flow of hypothesis testing can be summarized briefly as follows.

First, formulate the null and alternative hypotheses . Next, Calculate the p-value assuming that the null hypothesis is true. . Finally, Compare the p-value with the significance level and decide whether to reject the null hypothesis. 。

Through these three stages, researchers can use sample data to examine hypotheses about a population.

Important Principles in Hypothesis Testing

In hypothesis testing, it is important not merely to look at the numerical result but also to understand what was set as the null hypothesis、 what the p-value means、 what it means when the null hypothesis cannot be rejected It is important to understand this correctly.

A common beginner’s misunderstanding is to think that a small p-value means “the alternative hypothesis is highly likely to be true.” In fact, however, the p-value represents how unusual the data would be if the null hypothesis were assumed to be true It does not directly express the probability that either hypothesis itself is true or false.

Summary

Statistical hypothesis testing is a method for probabilistically evaluating a hypothesis about a population on the basis of sample data.

First, null hypothesis and alternative hypothesis First, the hypotheses are formulated. Next, assuming the null hypothesis is true, the p-value p-value—the probability of obtaining the observed data or data even more extreme—is calculated. If the p-value is at or below the significance level, the null hypothesis is rejected and the alternative hypothesis is supported.

However, Failing to reject the null hypothesis does not mean accepting it as true. This point requires careful attention.

Hypothesis testing is one of the central concepts of statistics and is an important tool for objectively evaluating the results of research, surveys, and experiments.