A researcher in planning research will develop a hypothesis from the research that will be conducted. The hypothesis is created as a proposition about the population parameters to be tested statistically through samples taken from the population.

Therefore, to determine the conclusion of a study, it is necessary to do statistical hypothesis testing. The statistical hypothesis consists of the null hypothesis (Ho) and the alternative hypothesis (Ha). The null hypothesis is a neutral statement with the sign “=”, while the alternative hypothesis is the opposite of the null hypothesis. Thus, if the null hypothesis is rejected in statistical hypothesis testing, then the alternative hypothesis is accepted and vice versa.

The statistical hypothesis tests the null hypothesis. The determination of whether the null hypothesis is accepted or rejected is carried out based on the purpose of testing the hypothesis. The criteria for accepting or rejecting the null hypothesis are based on the p-value. To determine whether the null hypothesis is accepted or rejected, it refers to the alpha value set in the study.

Generally, the researcher determines the alpha at 1% and 5% for experimental research, while for survey research, it can be set with an alpha value of up to 10%. Because of the importance of knowledge about hypothesis testing based on alpha values, on this occasion, Kanda Data will write about How to Distinguish 0.01, 0.05, and 0.10 Significance Levels in Statistics.

**How to write statistical hypotheses**

Before the researcher conducts the research, it will be preceded by formulating the null hypothesis and alternative hypotheses. In the formulation of hypotheses, it is known that there are one-tailed hypotheses and two-tailed hypotheses. The writing of these two types of hypotheses is different from one another.

Researchers need to write mathematical equations to facilitate the writing of statistical hypotheses. The one-tailed hypothesis in the mathematical equation contains signs “> and <.” On the other hand, there are signs for the two-tailed hypothesis: “= and ≠. Researchers can choose to use a one-tailed or two-tailed hypothesis according to the research objectives.

Researchers should use a two-tailed hypothesis test for research purposes that are not known for certain whether the effect is negative or positive. Meanwhile, if the researcher can ascertain the direction of the effect, the researcher can choose the one-tailed hypothesis.

To make it easier to understand writing statistical hypotheses, I will give an example of a case study. Suppose a researcher is conducting an observation to determine the difference in rice production before and after introducing new technology.

Researchers observed the average value of rice production before the introduction and then introduced the new technology for six months. Researchers observed the average rice production again after the introduction of new technology was completed. Based on this case example, researchers do not yet know whether the direction is positive or negative. Therefore, the researcher decided to use a two-tailed hypothesis.

Writing a mathematical hypothesis can be written as follows:

Ho: µ = µ 0 there is no significant difference between rice production before the introduction of technology and rice production after the introduction of new technology.

H1: µ ≠ µ 0 there is a significant difference between rice production before the introduction of technology and rice production after the introduction of new technology.

The writing of other hypotheses is adjusted to the researcher’s chosen analysis tool. Although, in principle, the same between the test of influence, correlation, and test of difference, it is better to use a different notation.

**Understanding the difference in p-value 0.01, 0.05, and 0.10**

Based on what I wrote in the previous paragraph, the researcher can determine the alpha at 1%, 5% or 10%. If the researcher determines an alpha of 5%, it can be analogous that out of 100 trials, failures are less than or equal to 5 times, then the study is declared a success. The same thing can also be analogized if the alpha is determined by 10%, meaning that if the success is 90 times out of 100 trials, the research is declared successful.

In studies with an environment that we can control well, researchers may consider using 5% or 1%. Alpha 5% and 1% can be applied to experimental studies with a relatively controllable research environment.

Especially for research in the medical field, it would be better if the alpha was determined to be smaller, for example, by 1%. On the other hand, survey studies with relatively difficult-to-control environments can determine an alpha of 10%. Therefore, researchers can choose to determine alpha according to their respective fields of study. The smaller the alpha value indicates, the higher the confidence level in a study.

**Statistical hypothesis test criteria**

After you understand the difference in alpha levels used in the study, you must have an in-depth understanding of the basic criteria for acceptance of the hypothesis. Following what I wrote in the previous paragraph, what is being tested is the null hypothesis.

Suppose we use the example case that I have written in the paragraph above for the test criteria, namely:

if the p-value > 0.05, then the null hypothesis is accepted

if the p-value ≤ 0.05, then the null hypothesis is rejected (the alternative hypothesis is accepted)

For example, based on the results of the t-test, if the p-value is 0.015, it indicates that the p-value is <0.05. Therefore, based on the criteria for acceptance of the hypothesis, it was decided that Ho was rejected. Because Ho is rejected, we accept the alternative hypothesis, namely that there is a significant difference between rice production before and rice production after introducing new technology.

However, if the alpha determined by the researcher is 1%, the null hypothesis is rejected because the p-value is > 0.01. Researchers can choose to test the hypothesis acceptance criteria following the previously determined alpha. That’s all I can write for all of you. Hopefully, it’s useful. Wait for the update of the Kanda data article next week!

Pingback: How to Find P-Value and T-Distribution Table using Excel - KANDA DATA

Pingback: How to Interpret Negative Coefficient Estimations in Linear Regression? - KANDA DATA