In inferential statistics, we need to formulate research hypotheses. These research hypotheses are formulated according to the research objectives. Furthermore, statistical hypotheses need to be established in the analysis method, consisting of null and alternative hypotheses.

Multiple linear regression analysis is a frequently utilized analytical tool for researchers to observe the influence of variables on one another. So, how do we conduct hypothesis testing in the context of multiple linear regression analysis? Let’s delve into and discuss this together in this article.

Multiple linear regression analysis is a tool we can employ to understand the influence of two or more independent variables on a single dependent variable. Based on this, we can conclude that multiple linear regression involves at least two independent variables.

In multiple linear regression, we can determine the simultaneous impact of independent variables on the dependent variable. Additionally, we can assess the partial influence of independent variables on the dependent variable.

In the analysis of multiple linear regression, the method commonly used is the Ordinary Least Squares (OLS) or least squares method. Naturally, the expected estimation results should be consistent and unbiased. Therefore, we need to perform the required assumption tests.

Assumption tests that need to be conducted for cross-sectional data include tests for residual normality, heteroskedasticity, multicollinearity, and linearity. If using time series data, in addition to these four assumption tests, an autocorrelation test needs to be added. These assumption tests aim to produce the Best Linear Unbiased Estimator (BLUE).

**Regression Output for Hypothesis Testing**

In the research methodology, it’s necessary to formulate statistical hypotheses about the combined effects of independent variables and the partial effects of independent variables on the dependent variable.

As previously mentioned, in the context of multiple linear regression analysis, we can examine the computed F-value if we want to test the combined effects of independent variables on the dependent variable. This F-value can also be used to assess the goodness of fit of the regression model we’ve developed.

Furthermore, we can observe the computed T-value to evaluate the partial impact of independent variables on the dependent variable.

**Hypothesis Testing Using Regression F-Test**

In the output of multiple linear regression analysis, the computed F-value is found in the ANOVA table. To assess the quality of our model, we can first look at the coefficient of determination (R-squared). The coefficient of determination explains the ability of the variation in the independent variables to explain the variation in the dependent variable.

Based on the computed F-value, we can compare it with the critical F-value from the F-table to conclude whether the combined effects of independent variables on the dependent variable are significant or not.

Furthermore, if we’re using data analysis software, we can directly use the significance value (p-value) to decide whether to accept or reject the null hypothesis.

**Hypothesis Testing Using Regression T-Test**

To determine the partial impact of independent variables on the dependent variable, we can use the regression T-test, often referred to as the t-statistics. Here, it’s important to note that the regression T-test differs from the t-test used for testing differences.

When testing the computed T-value, we have two criteria. We can either compare the T-value with the critical T-value from the T-table or examine the alpha P-value. Both approaches lead to the same conclusion.

**Formulating Regression Statistical Hypotheses**

We need to establish null and alternative hypotheses to facilitate hypothesis testing as mentioned in the previous paragraph.

The null hypothesis is often abbreviated as Ho, and the alternative hypothesis is abbreviated as H1 or Ha. The formulation of these two hypotheses is in contrast to each other. For example:

Ho: The combined effect of the amount of chemical fertilizer and the amount of compost fertilizer has no significant impact on rice production.

Ha: The combined effect of the amount of chemical fertilizer and the amount of compost fertilizer has a significant impact on rice production.

Ho: The partial effect of the amount of chemical fertilizer has no significant impact on rice production.

Ha: The partial effect of the amount of chemical fertilizer has a significant impact on rice production.

Ho: The partial effect of the amount of compost fertilizer has no significant impact on rice production.

Ha: The partial effect of the amount of compost fertilizer has a significant impact on rice production.

When writing statistical hypotheses, it’s essential to understand that when assessing the partial impact of independent variables, one of the variables is assumed to be held constant or ceteris paribus.

**Criteria for Null Hypothesis Testing**

Hypothesis testing criteria can utilize two methods: comparing the computed F-value/T-value with the critical F-table/T-table value and examining the alpha P-value concerning the research’s predetermined significance level.

It’s important to note that in hypothesis testing, the null hypothesis is the one being tested. When the null hypothesis is rejected, we accept the alternative hypothesis. Here are examples of criteria for hypothesis testing:

For F-test:

If computed F-value < F critical value: Null hypothesis is accepted.

If computed F-value ≥ F critical value: Null hypothesis is rejected (alternative hypothesis is accepted).

For T-test:

If computed T-value < T critical value: Null hypothesis is accepted.

If computed T-value ≥ T critical value: Null hypothesis is rejected (alternative hypothesis is accepted).

For P-value:

If P-value for F-test > 0.05: Null hypothesis is accepted.

If P-value for F-test ≤ 0.05: Null hypothesis is rejected (alternative hypothesis is accepted).

If P-value for T-test > 0.05: Null hypothesis is accepted.

If P-value for T-test ≤ 0.05: Null hypothesis is rejected (alternative hypothesis is accepted).

**Conclusion**

Well, now it’s time to conclude the article we’ve written in this instance. Similar to other forms of analysis, multiple linear regression analysis falls under inferential statistics, requiring us to conduct statistical hypothesis testing.

Statistical hypothesis testing is derived from research hypotheses to address the research objectives. Hypothesis testing in multiple linear regression analysis can be carried out using two methods: comparing the computed F-value/T-value with the critical F-table/T-table values or comparing the P-value with the predetermined significance level alpha established in the research.

In the context of statistical hypothesis testing, the null hypothesis is being examined. Thus, we accept the alternative hypothesis when the null hypothesis is rejected. Typically, research hypotheses are formulated as alternative hypotheses.

Well, that concludes the article I can provide on this occasion. I hope it’s beneficial and adds value to our collective knowledge. See you in the next week’s educational article. Thank you!