91Ó°ÊÓ

In hypothesis testing, examining paired differences involves comparing two related groups to determine any significant change or effect. This method is commonly used in before-and-after studies.
By analyzing the differences between paired observations, researchers can control for variability that might affect the results if examined separately. Paired difference tests are especially useful: - When the same subjects are measured under different conditions (e.g., performance before and after a training session). - In matching cases with controls. The main goal is to check if these paired differences arise purely by chance. If the population of these differences follows a normal distribution, then the subsequent steps using the t-distribution become valid.

The t-distribution is a probability distribution that is used when the sample size is small, or when the population variance is unknown. It resembles the normal distribution but has thicker tails. These fat tails occur because there is a greater probability of extreme values when dealing with smaller samples. Key facts about the t-distribution: - It is symmetric and bell-shaped, like the normal distribution. - As the sample size increases, the t-distribution approaches the normal distribution. - It is defined by degrees of freedom, calculated as the sample size minus one (df = n - 1). In hypothesis testing involving paired differences, the t-distribution allows for the calculation of critical values, giving us a reference for deciding if a test statistic is extreme enough to reject the null hypothesis.

The test statistic is a calculable value used to determine whether to reject the null hypothesis. For paired differences, the test statistic is based on the t-formula: \[ t = \frac{\bar{d} - \mu_d}{SE} \]Where:

• \(\bar{d}\) is the sample mean of the differences.
• \(\mu_d\) is the hypothesized mean difference, often set to zero.
• \(SE\) is the standard error, \(\frac{s_d}{\sqrt{n}}\).

This statistic quantifies how far the actual sample mean is from the hypothesized population mean in standard error units.
A larger test statistic in absolute value indicates evidence against the null hypothesis.

The critical value is a threshold that the test statistic has to exceed for us to reject the null hypothesis. It is determined based on the significance level \(\alpha\), the degree of confidence (or confidence level), and the specific t-distribution.To find the critical value:

• Use the t-distribution table or a statistical software.
• Consider the degrees of freedom (df = n - 1).
• Account for the type of test: one-tailed or two-tailed. For a two-tailed test, divide the significance level \(\alpha\) by 2.

When the test statistic is more extreme than this critical value (i.e., falls within the critical region), the evidence suggests that the null hypothesis should be rejected.