# Hypothesis Test for Correlation

## Understanding the Correlation Coefficient and Conducting a Hypothesis Test for Correlation

When examining bivariate data, one can determine the strength of the linear correlation between two variables using a correlation coefficient. This coefficient, also known as the product moment correlation coefficient (PMCC), ranges from -1 to 1 and expresses the relationship between the variables. A value of 1 represents perfect positive linear correlation, while a value of -1 represents perfect negative linear correlation. It's important to note that while correlation does not imply causation, a PMCC close to 1 or -1 suggests a higher likelihood of a relationship between the variables. Additionally, understanding the context of the data is crucial in accurately interpreting the results of the correlation coefficient.

In order to calculate the PMCC, one can use a graphics calculator or the formula provided in the formula booklet. The resulting value of r indicates the strength of the correlation, with values closer to 1 or -1 indicating a stronger correlation between the variables.

To determine the existence of a linear relationship between two variables in a population, a hypothesis test must be conducted. This involves comparing the calculated PMCC to a critical value. Before conducting the hypothesis test, it is important to understand the key terms: null hypothesis, alternative hypothesis, and test statistic. The null hypothesis ( ) is the initial assumption until proven otherwise, while the alternative hypothesis ( ) is the conclusion reached if the null hypothesis is rejected. The test statistic, calculated from the sample, is then compared to the critical value in the significance test. The critical region determines the values that lead to the rejection of the null hypothesis, while the significance level represents the probability of rejecting the null hypothesis when it is actually true.

## Hypothesis Test for Negative Correlation

The hypothesis test method involves the following steps:

• 1. Write down the null and alternative hypotheses ( and ). The null hypothesis is always , while the alternative hypothesis depends on the question at hand. Both hypotheses should be stated in symbols only, not in words.
• 2. Use a calculator to calculate the PMCC (r) of the sample data.
• 3. Determine the critical value by referencing the PMCC table in the formula booklet and taking into account the significance level and sample size.
• 4. Compare the absolute values of the PMCC and r to the critical value. If the absolute value is greater, the null hypothesis should be rejected. Otherwise, the null hypothesis should be accepted.
• 5. Write a comprehensive conclusion that includes both statistical language and the context of the question.

By following these steps, a hypothesis test for correlation can be effectively conducted to determine the presence of a linear relationship between two variables. It is important to consider both the null and alternative hypotheses, as well as the critical value and significance level, when interpreting the results of the test. Proper understanding of the correlation coefficient and null hypothesis is essential in accurately assessing the relationship between variables and drawing meaningful conclusions from bivariate data.

## The Role of the Null Hypothesis in Correlation Testing

In correlation testing, the null hypothesis states that there is no relationship between the two variables being studied. On the other hand, the alternative hypothesis suggests that there is a correlation between the two variables.

To interpret the results based on the null hypothesis, a decision must be made using the observed results, also known as the test statistic. This decision determines whether to reject the null hypothesis or not. A significance level of 5% is often used in both one-tailed and two-tailed tests, meaning there is a 5% chance of making a false conclusion.

The distinction between one-tailed and two-tailed tests lies in the distribution of the significance level. In a one-tailed test, the significance level of 5% is allocated on one side, while in a two-tailed test, it is evenly distributed on both sides. In order for the observed result to be considered statistically significant, it must fall within the 5% significance level range.

When the null hypothesis is rejected, the alternative hypothesis is utilized to draw a conclusion. This is done by stating the null and alternative hypotheses for both one-tailed and two-tailed tests.

For a one-tailed test:
Null hypothesis: ρ = 0
Alternative hypothesis: ρ > 0

For a two-tailed test:
Null hypothesis: ρ = 0
Alternative hypothesis: ρ ≠ 0

To better understand how correlation testing works, let's consider an example. A group of 12 students took both a theoretical and practical biology test, with their scores shown in the table below:

• Theoretical test scores: 5, 9, 7, 11, 20, 46, 17, 12, 10, 15, 16
• Practical test scores: 6, 8, 9, 13, 20, 98, 17, 14, 18, 17, 18

a) To find the product moment correlation coefficient for this data, we can use a calculator to calculate the PMCC by entering the data into two lists and calculating the regression line. The PMCC is found to be r = 0.935, with 3 significant figures.

b) A teacher claims that students who excel in the theoretical test also tend to do well in the practical test. To test this claim at a significance level of 0.05, we must clearly state the null and alternative hypotheses. As the claim is for a positive correlation, the following hypotheses are used:

Null hypothesis: ρ = 0
Alternative hypothesis: ρ > 0

The five steps for hypothesis testing can now be applied:

1. State the null and alternative hypotheses.The null hypothesis is ρ = 0, while the alternative hypothesis is ρ > 0.

2. Calculate the PMCC.Using the calculator, we find that the PMCC is r = 0.935.

3. Determine the critical value.For a one-tailed test with a significance level of 5% and a sample size of 12, the critical value is shown to be cv = 0.4973 in the formula booklet.

4. Compare the PMCC to the critical value.Since the absolute value of the PMCC (0.935) is greater than the critical value (0.4973), the null hypothesis is rejected and the alternative hypothesis is accepted.

5. Draw a conclusion.Based on the observed result, it can be concluded that there is significant evidence to support the claim that students who do well in the theoretical biology test also tend to do well in the practical biology test.

Let's examine another example to further understand the process of hypothesis testing for correlation.

A tetrahedral die (four faces) is rolled 40 times, with 6 'ones' being observed. Is there any evidence at a significance level of 0.10 that the probability of getting a 'one' is less than a quarter?

To test this claim, the same five steps are followed:

1. State the null and alternative hypotheses.Null hypothesis: ρ = 0
Alternative hypothesis: ρ < 0.25

2. Calculate the PMCC.As only the frequency of 'ones' is given, the PMCC cannot be calculated.

3. Determine the critical value.Since a one-tailed test is needed at a 10% significance level, it is converted to a binomial distribution and the cumulative binomial tables are used. The observed value is 6.

4. Compare the observed result to the critical value.The critical value is found to be 0.0962, or 9.62%. As this value is less than 10%, the observed result falls within the critical region.

5. Draw a conclusion.Based on the observed result, the null hypothesis is rejected and the alternative hypothesis is accepted. This suggests evidence that the probability of getting a 'one' is less than a quarter.

## Understanding Hypothesis Testing for Correlation

When analyzing data, it is important to determine if there is a relationship between two variables. The Pearson correlation coefficient, also known as r value, is a measure of this relationship. It ranges from -1 to 1, with higher values indicating a stronger positive or negative correlation. However, is this correlation enough to draw conclusions? Let's take a closer look at hypothesis testing for correlation.

First, it's important to note that correlation does not equate to causation. Just because two variables are correlated does not mean that one causes the other. Therefore, it is necessary to use hypothesis testing to determine if the correlation is significant or just due to chance.

To conduct a hypothesis test with correlation, we must establish a null and alternative hypothesis. The null hypothesis assumes that there is no correlation between the variables, while the alternative hypothesis is accepted if the null hypothesis is rejected. This can be done through a one-tailed test, focusing on either a positive or negative correlation, or a two-tailed test, considering both options.

Next, we gather sample data and calculate the r value. Then, using a significance level and sample size, we can find the critical value from a table. By comparing the r value to the critical value, we can determine if the alternative hypothesis is accepted. For instance, if the r value is greater than the critical value, we can accept the alternative hypothesis with a 95% confidence level.

In conclusion, hypothesis testing for correlation is a useful tool in statistical analysis. By establishing a null and alternative hypothesis and utilizing the five-step process, we can make meaningful conclusions and accurate predictions. The Pearson correlation coefficient, with its range of -1 to 1, reveals the strength of a relationship between two variables. Keep in mind, however, that correlation alone does not prove causation.